renderer: emulate clip-disabled screen-space positions

This reverts commit 6758d67942.

An EXEC pair loses its mask provenance and falls into the raw high-word path. The saved all-ones high word then keeps waterfall branches true and wedges GPU work.

CMakeLists.txt

@@ -574,6 +574,8 @@ if(BUILD_TESTING)
 		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --storage-sampled-only)
 	add_test(NAME texture_cache_depth_readback
 		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --depth-readback-only)
+	add_test(NAME compute_meta_clear_classification
+		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --compute-meta-clear-only)
 	add_test(NAME buffer_cache_dirty_gc
 		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --buffer-cache-gc-only)
 

src/graphics/guest_gpu/command_processor/pm4Dispatch.cpp

@@ -23,9 +23,8 @@ constexpr auto MakeContextDispatchTable() {
 	g_hw_ctx_func[Pm4::DB_STENCIL_CLEAR]             = HwCtxSetStencilClear;
 	g_hw_ctx_func[Pm4::DB_DEPTH_CLEAR]               = HwCtxSetDepthClear;
 	g_hw_ctx_func[Pm4::PA_SC_SCREEN_SCISSOR_TL]      = HwCtxSetScreenScissor;
-	g_hw_ctx_func[Pm4::DB_Z_INFO]                    = HwCtxSetDepthRenderTarget;
 	g_hw_ctx_func[Pm4::DB_STENCIL_INFO]              = HwCtxSetStencilInfo;
-	g_hw_ctx_func[Pm4::DB_HTILE_SURFACE]             = HwCtxSetDepthHtileSurface;
+	g_hw_ctx_func[Pm4::DB_SHADING_RATE_ENCODING]     = HwCtxSetDepthShadingRateEncoding;
 	g_hw_ctx_func[Pm4::PA_SU_HARDWARE_SCREEN_OFFSET] = HwCtxSetHardwareScreenOffset;
 	g_hw_ctx_func[Pm4::PA_SC_WINDOW_OFFSET]          = HwCtxSetWindowOffset;
 	g_hw_ctx_func[Pm4::PA_SC_WINDOW_SCISSOR_TL]      = HwCtxSetWindowScissor;
@@ -75,7 +74,6 @@ constexpr auto MakeContextDispatchTable() {
 	g_hw_ctx_func[Pm4::PA_SC_BINNER_CNTL_0]                   = HwCtxSetPaScExtendedControl;
 	g_hw_ctx_func[Pm4::PA_SC_BINNER_CNTL_1]                   = HwCtxSetPaScExtendedControl;
 	g_hw_ctx_func[Pm4::PA_SC_CONSERVATIVE_RASTERIZATION_CNTL] = HwCtxSetPaScExtendedControl;
-	g_hw_ctx_func[Pm4::PA_SC_NGG_MODE_CNTL]                   = HwCtxSetPaScExtendedControl;
 	g_hw_ctx_func[Pm4::DB_ALPHA_TO_MASK]                      = HwCtxSetAlphaToMask;
 	g_hw_ctx_func[Pm4::VGT_DRAW_PAYLOAD_CNTL]                 = HwCtxSetDrawPayloadControl;
 	g_hw_ctx_func[Pm4::VGT_PRIMITIVEID_RESET]                 = HwCtxSetPrimitiveIdReset;
@@ -83,7 +81,6 @@ constexpr auto MakeContextDispatchTable() {
 	g_hw_ctx_func[Pm4::VGT_SHADER_STAGES_EN]                  = HwCtxSetShaderStages;
 
 	for (uint32_t slot = 0; slot < 8; slot++) {
-		g_hw_ctx_func[Pm4::CB_COLOR0_BASE + slot * 15] = HwCtxSetRenderTarget;
 		g_hw_ctx_func[Pm4::CB_COLOR0_INFO + slot * 15] = HwCtxSetColorInfo;
 
 		g_hw_ctx_func[Pm4::CB_BLEND0_CONTROL + slot * 1] = HwCtxSetBlendControl;
@@ -116,25 +113,21 @@ constexpr auto MakeShaderDispatchTable() {
 	}
 
 	for (uint32_t slot = 0; slot < 4; slot++) {
-		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_PS_0 + slot] = HwShSetPsUserAccumSgpr;
+		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_PS_0 + slot] = HwShIgnoreUserAccumulator;
 	}
 
 	for (uint32_t slot = 0; slot < 4; slot++) {
-		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_ESGS_0 + slot] = HwShSetGsUserAccumSgpr;
+		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_ESGS_0 + slot] = HwShIgnoreUserAccumulator;
 	}
 	for (uint32_t slot = 0; slot < 4; slot++) {
-		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_LSHS_0 + slot] = HwShSetHsUserAccumSgpr;
+		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_LSHS_0 + slot] = HwShIgnoreUserAccumulator;
 	}
 
-	for (uint32_t slot = 0; Pm4::COMPUTE_USER_ACCUM_0 + slot < Pm4::COMPUTE_USER_DATA_0; slot++) {
-		g_hw_sh_func[Pm4::COMPUTE_USER_ACCUM_0 + slot] = HwShSetCsUserAccumSgpr;
+	for (uint32_t slot = 0; slot < 4; slot++) {
+		g_hw_sh_func[Pm4::COMPUTE_USER_ACCUM_0 + slot] = HwShIgnoreUserAccumulator;
 	}
 	g_hw_sh_func[Pm4::COMPUTE_PGM_LO]                  = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_PGM_HI]                  = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_TBA_LO]                  = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_TBA_HI]                  = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_TMA_LO]                  = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_TMA_HI]                  = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_PGM_RSRC1]               = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_PGM_RSRC2]               = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_START_X]                 = HwShSetCsRegisters;
@@ -144,30 +137,16 @@ constexpr auto MakeShaderDispatchTable() {
 	g_hw_sh_func[Pm4::COMPUTE_NUM_THREAD_Y]            = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_NUM_THREAD_Z]            = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_RESOURCE_LIMITS]         = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_DESTINATION_EN_SE0]      = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_DESTINATION_EN_SE1]      = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_TMPRING_SIZE]            = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_DESTINATION_EN_SE2]      = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_DESTINATION_EN_SE3]      = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_PGM_RSRC3]               = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_SHADER_CHKSUM]           = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_DISPATCH_TUNNEL]         = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_TBA_LO_PS]            = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC3_PS]         = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC4_PS]         = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_TMA_LO_PS]            = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_TMA_HI_PS]            = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_REQ_CTRL_PS]          = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC3_GS]         = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC4_GS]         = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::COMPUTE_PACE_ID]                 = HwShSetCsRegisters;
+	g_hw_sh_func[Pm4::SPI_GRAPHICS_SHADER_CONTROL_PS]  = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::SPI_GRAPHICS_SHADER_CONTROL_GS]  = HwShIgnoreRegisters;
 	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_LO_GS] = HwShIgnoreRegisters;
 	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_HI_GS] = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_REQ_CTRL_ESGS]        = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC3_HS]         = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC4_HS]         = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::SPI_GRAPHICS_SHADER_CONTROL_HS]  = HwShIgnoreRegisters;
 	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_LO_HS] = HwShIgnoreRegisters;
 	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_HI_HS] = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_REQ_CTRL_LSHS]        = HwShIgnoreRegisters;
 
 	return g_hw_sh_func;
 }
@@ -179,6 +158,7 @@ constexpr auto MakeUconfigDispatchTable() {
 	g_hw_uc_func[Pm4::VGT_INDEX_TYPE]            = HwUcSetIndexType;
 	g_hw_uc_func[Pm4::VGT_OBJECT_ID]             = HwUcSetObjectId;
 	g_hw_uc_func[Pm4::TEXTURE_GRADIENT_FACTORS]  = HwUcSetTextureGradientFactors;
+	g_hw_uc_func[Pm4::TEXTURE_GRADIENT_CONTROL]  = HwUcSetTextureGradientFactors;
 	g_hw_uc_func[Pm4::IA_MULTI_VGT_PARAM]        = HwUcSetIaMultiVgtParam;
 	g_hw_uc_func[Pm4::GE_MULTI_PRIM_IB_RESET_EN] = HwUcSetMultiPrimIbReset;
 	g_hw_uc_func[Pm4::TA_CS_BC_BASE_ADDR]        = HwUcSetBorderColorTableAddr;

src/graphics/guest_gpu/command_processor/pm4Dispatch.h

@@ -44,11 +44,9 @@ uint32_t HwCtxSetBorderColorTableAddr(CommandProcessor&, uint32_t, uint32_t, con
 uint32_t HwCtxSetStencilClear(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetDepthClear(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetScreenScissor(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwCtxSetDepthRenderTarget(CommandProcessor&, uint32_t, uint32_t, const uint32_t*,
-                                   uint32_t);
 uint32_t HwCtxSetStencilInfo(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwCtxSetDepthHtileSurface(CommandProcessor&, uint32_t, uint32_t, const uint32_t*,
-                                   uint32_t);
+uint32_t HwCtxSetDepthShadingRateEncoding(CommandProcessor&, uint32_t, uint32_t,
+                                          const uint32_t*, uint32_t);
 uint32_t HwCtxSetHardwareScreenOffset(CommandProcessor&, uint32_t, uint32_t, const uint32_t*,
                                       uint32_t);
 uint32_t HwCtxSetWindowOffset(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
@@ -89,7 +87,6 @@ uint32_t HwCtxSetDrawPayloadControl(CommandProcessor&, uint32_t, uint32_t, const
 uint32_t HwCtxSetPrimitiveIdReset(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetObjprimIdControl(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetShaderStages(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwCtxSetRenderTarget(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetColorInfo(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetBlendControl(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetViewportScissor(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
@@ -100,10 +97,8 @@ uint32_t HwShSetPsUserSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t
 uint32_t HwShSetGsUserSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwShSetHsUserSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwShSetCsUserSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwShSetPsUserAccumSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwShSetGsUserAccumSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwShSetHsUserAccumSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwShSetCsUserAccumSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
+uint32_t HwShIgnoreUserAccumulator(CommandProcessor&, uint32_t, uint32_t, const uint32_t*,
+                                   uint32_t);
 uint32_t HwShSetCsRegisters(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwShIgnoreRegisters(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwUcSetPrimitiveType(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);

src/graphics/guest_gpu/gpu_defs.h

@@ -407,7 +407,19 @@ enum class BufferFormat : uint32_t {
 	k5_5_5_1UNorm       = 134,
 	k1_5_5_5UNorm       = 135,
 	k4_4_4_4UNorm       = 136,
+	kFmask8_S2_F1       = 156,
+	kFmask8_S4_F1       = 157,
+	kFmask8_S8_F1       = 158,
+	kFmask8_S2_F2       = 159,
+	kFmask8_S4_F2       = 160,
 	kFmask8_S4_F4       = 161,
+	kFmask16_S16_F1     = 162,
+	kFmask16_S8_F2      = 163,
+	kFmask32_S16_F2     = 164,
+	kFmask32_S8_F4      = 165,
+	kFmask32_S8_F8      = 166,
+	kFmask64_S16_F4     = 167,
+	kFmask64_S16_F8     = 168,
 	kBc1UNorm           = 169,
 	kBc1Srgb            = 170,
 	kBc2UNorm           = 171,

src/graphics/guest_gpu/gpu_format.cpp

@@ -68,7 +68,19 @@ constexpr FormatInfo kFormatInfo[] = {
 	{BufferFormat::k5_5_5_1UNorm, 2, 0, 2, true, false},
 	{BufferFormat::k1_5_5_5UNorm, 0, 0, 2, false, false},
 	{BufferFormat::k4_4_4_4UNorm, 2, 0, 2, true, false},
+	{BufferFormat::kFmask8_S2_F1, 1, 0, 1, true, false},
+	{BufferFormat::kFmask8_S4_F1, 1, 0, 1, true, false},
+	{BufferFormat::kFmask8_S8_F1, 1, 0, 1, true, false},
+	{BufferFormat::kFmask8_S2_F2, 1, 0, 1, true, false},
+	{BufferFormat::kFmask8_S4_F2, 1, 0, 1, true, false},
 	{BufferFormat::kFmask8_S4_F4, 1, 0, 1, true, false},
+	{BufferFormat::kFmask16_S16_F1, 2, 0, 2, true, false},
+	{BufferFormat::kFmask16_S8_F2, 2, 0, 2, true, false},
+	{BufferFormat::kFmask32_S16_F2, 4, 0, 4, true, false},
+	{BufferFormat::kFmask32_S8_F4, 4, 0, 4, true, false},
+	{BufferFormat::kFmask32_S8_F8, 4, 0, 4, true, false},
+	{BufferFormat::kFmask64_S16_F4, 8, 0, 8, true, false},
+	{BufferFormat::kFmask64_S16_F8, 8, 0, 8, true, false},
 	{BufferFormat::kBc1UNorm, 0, 8, 0, true, false},
 	{BufferFormat::kBc1Srgb, 0, 8, 0, true, false},
 	{BufferFormat::kBc2UNorm, 0, 16, 0, true, false},
@@ -203,7 +215,7 @@ uint32_t RenderTargetBytesPerElement(BufferFormat format) {
 }
 
 bool IsFmaskTextureFormat(BufferFormat format) {
-	return format == BufferFormat::kFmask8_S4_F4;
+	return format >= BufferFormat::kFmask8_S2_F1 && format <= BufferFormat::kFmask64_S16_F8;
 }
 
 bool IsSampledTextureFormat(BufferFormat format) {

src/graphics/guest_gpu/hardwareContext.h

@@ -11,15 +11,6 @@ struct ColorBase {
 	uint64_t addr = 0;
 };
 
-struct ColorPitch {
-	uint32_t pitch_div8_minus1       = 0;
-	uint32_t fmask_pitch_div8_minus1 = 0;
-};
-
-struct ColorSlice {
-	uint32_t slice_div64_minus1 = 0;
-};
-
 struct ColorView {
 	uint32_t base_array_slice_index = 0;
 	uint32_t last_array_slice_index = 0;
@@ -36,20 +27,15 @@ struct ColorInfo {
 	bool                    blend_clamp                    = false;
 	bool                    blend_bypass                   = false;
 	bool                    round_mode                     = false;
-	uint32_t                cmask_is_linear                = 0;
-	uint32_t                cmask_addr_type                = 0;
-	bool                    alt_tile_mode                  = false;
 	Prospero::ChannelLayout format                         = Prospero::ChannelLayout::kInvalid;
 	Prospero::ChannelType   channel_type                   = Prospero::ChannelType::kUNorm;
 	Prospero::ChannelOrder  channel_order                  = Prospero::ChannelOrder::kStandard;
 };
 
 struct ColorAttrib {
-	bool               force_dest_alpha_to_one = false;
-	Prospero::TileMode tile_mode               = Prospero::TileMode::kLinear;
-	Prospero::TileMode fmask_tile_mode         = Prospero::TileMode::kLinear;
-	uint32_t           num_samples             = 0;
-	uint32_t           num_fragments           = 0;
+	bool     force_dest_alpha_to_one = false;
+	uint32_t num_samples             = 0;
+	uint32_t num_fragments           = 0;
 };
 
 struct ColorAttrib2 {
@@ -59,41 +45,33 @@ struct ColorAttrib2 {
 };
 
 struct ColorAttrib3 {
-	uint32_t           depth              = 0;
-	Prospero::TileMode tile_mode          = Prospero::TileMode::kLinear;
-	uint32_t           dimension          = 0;
-	bool               cmask_pipe_aligned = false;
-	bool               dcc_pipe_aligned   = false;
+	uint32_t           depth                        = 0;
+	Prospero::TileMode tile_mode                    = Prospero::TileMode::kRenderTarget;
+	uint32_t           dimension                    = 0;
+	bool               metadata_pipe_aligned        = false;
+	bool               write_vrs_rate_hint_to_cmask = false;
 };
 
 struct ColorDccControl {
-	uint32_t max_uncompressed_block_size    = 0;
-	uint32_t max_compressed_block_size      = 0;
-	uint32_t min_compressed_block_size      = 0;
-	uint32_t color_transform                = 0;
-	bool     dcc_clear_key_enable           = false;
-	bool     overwrite_combiner_disable     = false;
-	bool     independent_64b_blocks         = false;
-	bool     independent_128b_blocks        = false;
-	bool     data_write_on_dcc_clear_to_reg = false;
+	enum class IndependentBlockSize : uint8_t { Disabled, Bytes64, Bytes128 };
+
+	uint32_t             max_uncompressed_block_size    = 2;
+	uint32_t             max_compressed_block_size      = 2;
+	uint32_t             color_transform                = 0;
+	bool                 dcc_clear_key_enable           = false;
+	bool                 overwrite_combiner_disable     = false;
+	IndependentBlockSize independent_block_size         = IndependentBlockSize::Disabled;
+	bool                 data_write_on_dcc_clear_to_reg = false;
 };
 
 struct ColorCmask {
 	uint64_t addr = 0;
 };
 
-struct ColorCmaskSlice {
-	uint32_t slice_minus1 = 0;
-};
-
 struct ColorFmask {
 	uint64_t addr = 0;
 };
 
-struct ColorFmaskSlice {
-	uint32_t slice_minus1 = 0;
-};
-
 struct ColorClearWord0 {
 	uint32_t word0 = 0;
 };
@@ -106,15 +84,8 @@ struct ColorDccAddr {
 	uint64_t addr = 0;
 };
 
-struct ColorSize {
-	uint32_t width  = 0;
-	uint32_t height = 0;
-};
-
 struct RenderTarget {
 	ColorBase       base;
-	ColorPitch      pitch;
-	ColorSlice      slice;
 	ColorView       view;
 	ColorInfo       info;
 	ColorAttrib     attrib;
@@ -122,13 +93,10 @@ struct RenderTarget {
 	ColorAttrib3    attrib3;
 	ColorDccControl dcc;
 	ColorCmask      cmask;
-	ColorCmaskSlice cmask_slice;
 	ColorFmask      fmask;
-	ColorFmaskSlice fmask_slice;
 	ColorClearWord0 clear_word0;
 	ColorClearWord1 clear_word1;
 	ColorDccAddr    dcc_addr;
-	ColorSize       size;
 };
 
 struct DepthZInfo {
@@ -136,7 +104,7 @@ struct DepthZInfo {
 	uint32_t                                    num_samples = 0;
 	Prospero::TextureCompatiblePlaneCompression texture_compatibility =
 	    Prospero::TextureCompatiblePlaneCompression::kDisable;
-	Prospero::ZCompareBase z_compare_base     = Prospero::ZCompareBase::kZMin;
+	Prospero::ZCompareBase z_compare_base     = Prospero::ZCompareBase::kZMax;
 	bool                   htile_acceleration = false;
 	bool                   expclear_enabled   = false;
 	bool                   partially_resident = false;
@@ -171,7 +139,7 @@ struct DepthStencilInfo {
 	Prospero::TextureCompatibleStencil texture_compatibility =
 	    Prospero::TextureCompatibleStencil::kDisable;
 	bool expclear_enabled       = false;
-	bool htile_stencil_disabled = false;
+	bool htile_stencil_disabled = true;
 	bool partially_resident     = false;
 
 	[[nodiscard]] static DepthStencilInfo Decode(uint32_t value) {
@@ -194,16 +162,6 @@ struct DepthStencilInfo {
 	}
 };
 
-struct DepthRenderTargetDepthInfo {
-	uint32_t addr5_swizzle_mask = 0;
-	uint32_t array_mode         = 0;
-	uint32_t pipe_config        = 0;
-	uint32_t bank_width         = 0;
-	uint32_t bank_height        = 0;
-	uint32_t macro_tile_aspect  = 0;
-	uint32_t num_banks          = 0;
-};
-
 struct DepthDepthView {
 	uint32_t slice_start           = 0;
 	uint32_t slice_max             = 0;
@@ -218,36 +176,18 @@ struct DepthDepthSizeXY {
 	bool     valid = false;
 };
 
-struct DepthRenderTargetHTileSurface {
-	uint32_t linear                  = 0;
-	uint32_t full_cache              = 0;
-	uint32_t htile_uses_preload_win  = 0;
-	uint32_t preload                 = 0;
-	uint32_t prefetch_width          = 0;
-	uint32_t prefetch_height         = 0;
-	uint32_t dst_outside_zero_to_one = 0;
-};
-
 struct DepthRenderTarget {
-	DepthZInfo                    z_info;
-	DepthStencilInfo              stencil_info;
-	DepthDepthView                depth_view;
-	DepthDepthSizeXY              size;
-	DepthRenderTargetDepthInfo    depth_info;
-	DepthRenderTargetHTileSurface htile_surface;
+	DepthZInfo       z_info;
+	DepthStencilInfo stencil_info;
+	DepthDepthView   depth_view;
+	DepthDepthSizeXY size;
 
 	uint64_t z_read_base_addr        = 0;
 	uint64_t stencil_read_base_addr  = 0;
 	uint64_t z_write_base_addr       = 0;
 	uint64_t stencil_write_base_addr = 0;
-	uint32_t pitch_div8_minus1       = 0;
-	uint32_t height_div8_minus1      = 0;
-	bool     pitch_height_valid      = false;
-	uint32_t slice_div64_minus1      = 0;
 	uint64_t htile_data_base_addr    = 0;
-	uint32_t width                   = 0;
-	uint32_t height                  = 0;
-	bool     width_height_valid      = false;
+	uint8_t  shading_rate_encoding   = 0;
 };
 
 struct RenderControl {
@@ -256,6 +196,8 @@ struct RenderControl {
 	bool    resummarize_enable       = false;
 	bool    stencil_compress_disable = false;
 	bool    depth_compress_disable   = false;
+	bool    copy_depth_to_color      = false;
+	bool    copy_stencil_to_color    = false;
 	bool    copy_centroid            = false;
 	uint8_t copy_sample              = 0;
 };
@@ -299,16 +241,14 @@ struct ClipControl {
 };
 
 struct DepthControl {
-	bool    stencil_enable                     = false;
-	bool    z_enable                           = false;
-	bool    z_write_enable                     = false;
-	bool    depth_bounds_enable                = false;
-	uint8_t zfunc                              = 0;
-	bool    backface_enable                    = false;
-	uint8_t stencilfunc                        = 0;
-	uint8_t stencilfunc_bf                     = 0;
-	bool    color_writes_on_depth_fail_enable  = false;
-	bool    color_writes_on_depth_pass_disable = false;
+	bool    stencil_enable      = false;
+	bool    z_enable            = false;
+	bool    z_write_enable      = false;
+	bool    depth_bounds_enable = false;
+	uint8_t zfunc               = 0;
+	bool    backface_enable     = false;
+	uint8_t stencilfunc         = 0;
+	uint8_t stencilfunc_bf      = 0;
 };
 
 struct StencilControl {
@@ -468,27 +408,6 @@ struct ScreenViewport {
 	bool     clip_rect_window_offset_enable[4]    = {};
 };
 
-struct VsShaderResource1 {
-	uint8_t vgprs                    = 0;
-	uint8_t sgprs                    = 0;
-	uint8_t priority                 = 0;
-	uint8_t float_mode               = 0;
-	bool    dx10_clamp               = false;
-	bool    ieee_mode                = false;
-	uint8_t vgpr_component_count     = 0;
-	bool    cu_group_enable          = false;
-	bool    require_forward_progress = false;
-	bool    fp16_overflow            = false;
-};
-
-struct VsShaderResource2 {
-	bool    scratch_en        = false;
-	uint8_t user_sgpr         = 0;
-	bool    offchip_lds       = false;
-	bool    streamout_enabled = false;
-	uint8_t shared_vgprs      = 0;
-};
-
 struct HsShaderResource1 {
 	uint8_t vgprs                     = 0;
 	uint8_t priority                  = 0;
@@ -511,7 +430,6 @@ struct HsShaderResource2 {
 
 struct PsShaderResource1 {
 	uint8_t vgprs                    = 0;
-	uint8_t sgprs                    = 0;
 	uint8_t priority                 = 0;
 	uint8_t float_mode               = 0;
 	bool    dx10_clamp               = false;
@@ -535,7 +453,6 @@ struct PsStageRegisters {
 	uint64_t          data_addr = 0;
 	PsShaderResource1 rsrc1;
 	PsShaderResource2 rsrc2;
-	uint64_t          chksum = 0;
 };
 
 struct CsStageRegisters {
@@ -545,12 +462,13 @@ struct CsStageRegisters {
 	uint32_t num_thread_y              = 0;
 	uint32_t num_thread_z              = 0;
 	uint8_t  vgprs                     = 0;
-	uint8_t  sgprs                     = 0;
+	uint8_t  priority                  = 0;
 	uint8_t  float_mode                = 0;
 	bool     dx10_clamp                = false;
+	bool     debug_mode                = false;
 	bool     ieee_mode                 = false;
+	bool     require_forward_progress  = false;
 	bool     fp16_overflow             = false;
-	bool     bulky                     = false;
 	bool     threadgroup_configuration = false;
 	uint8_t  wave_size                 = 64;
 	bool     scratch_en                = false;
@@ -561,6 +479,7 @@ struct CsStageRegisters {
 	bool     tg_size_en                = false;
 	uint8_t  tidig_comp_cnt            = 0;
 	uint16_t lds_size                  = 0;
+	uint8_t  shared_vgprs              = 0;
 };
 
 struct EsStageRegisters {
@@ -568,21 +487,17 @@ struct EsStageRegisters {
 };
 
 struct LsStageRegisters {
-	uint64_t          data_addr = 0;
-	VsShaderResource1 rsrc1;
-	VsShaderResource2 rsrc2;
+	uint64_t data_addr = 0;
 };
 
 struct HsStageRegisters {
 	uint64_t          data_addr = 0;
 	HsShaderResource1 rsrc1;
 	HsShaderResource2 rsrc2;
-	uint64_t          chksum = 0;
 };
 
 struct GsShaderResource1 {
 	uint8_t vgprs                     = 0;
-	uint8_t sgprs                     = 0;
 	uint8_t priority                  = 0;
 	uint8_t float_mode                = 0;
 	bool    dx10_clamp                = false;
@@ -608,7 +523,6 @@ struct GsStageRegisters {
 	uint64_t          data_addr = 0;
 	GsShaderResource1 rsrc1;
 	GsShaderResource2 rsrc2;
-	uint64_t          chksum = 0;
 };
 
 struct ShaderRegisters {
@@ -664,8 +578,7 @@ struct ShaderRegisters {
 enum class UserSgprType { Unknown, Region, Vsharp };
 
 struct UserSgprInfo {
-	// Changed from 16 to 64
-	static constexpr int SGPRS_MAX = 64;
+	static constexpr int SGPRS_MAX = 32;
 
 	uint32_t     value[SGPRS_MAX] = {0};
 	UserSgprType type[SGPRS_MAX]  = {};
@@ -712,12 +625,6 @@ public:
 	void Reset() { *this = Context(); }
 
 	void SetColorBase(uint32_t slot, const ColorBase& base) { m_render_targets[slot].base = base; }
-	void SetColorPitch(uint32_t slot, const ColorPitch& pitch) {
-		m_render_targets[slot].pitch = pitch;
-	}
-	void SetColorSlice(uint32_t slot, const ColorSlice& slice) {
-		m_render_targets[slot].slice = slice;
-	}
 	void SetColorView(uint32_t slot, const ColorView& view) { m_render_targets[slot].view = view; }
 	void SetColorInfo(uint32_t slot, const ColorInfo& info) { m_render_targets[slot].info = info; }
 	void SetColorAttrib(uint32_t slot, const ColorAttrib& attrib) {
@@ -735,15 +642,9 @@ public:
 	void SetColorCmask(uint32_t slot, const ColorCmask& cmask) {
 		m_render_targets[slot].cmask = cmask;
 	}
-	void SetColorCmaskSlice(uint32_t slot, const ColorCmaskSlice& cmask_slice) {
-		m_render_targets[slot].cmask_slice = cmask_slice;
-	}
 	void SetColorFmask(uint32_t slot, const ColorFmask& fmask) {
 		m_render_targets[slot].fmask = fmask;
 	}
-	void SetColorFmaskSlice(uint32_t slot, const ColorFmaskSlice& fmask_slice) {
-		m_render_targets[slot].fmask_slice = fmask_slice;
-	}
 	void SetColorClearWord0(uint32_t slot, const ColorClearWord0& clear_word0) {
 		m_render_targets[slot].clear_word0 = clear_word0;
 	}
@@ -753,7 +654,6 @@ public:
 	void SetColorDccAddr(uint32_t slot, const ColorDccAddr& dcc_addr) {
 		m_render_targets[slot].dcc_addr = dcc_addr;
 	}
-	void SetColorSize(uint32_t slot, const ColorSize& size) { m_render_targets[slot].size = size; }
 	[[nodiscard]] const RenderTarget& GetRenderTarget(uint32_t slot) const {
 		return m_render_targets[slot];
 	}
@@ -1072,21 +972,16 @@ public:
 
 	void SetEsShaderBase(uint64_t addr) { m_vs.es_regs.data_addr = addr; }
 	void SetLsShaderBase(uint64_t addr) { m_vs.ls_regs.data_addr = addr; }
-	void SetLsShaderResource1(const VsShaderResource1& rsrc1) { m_vs.ls_regs.rsrc1 = rsrc1; }
-	void SetLsShaderResource2(const VsShaderResource2& rsrc2) { m_vs.ls_regs.rsrc2 = rsrc2; }
 	void SetHsShaderBase(uint64_t addr) { m_vs.hs_regs.data_addr = addr; }
 	void SetHsShaderResource1(const HsShaderResource1& rsrc1) { m_vs.hs_regs.rsrc1 = rsrc1; }
 	void SetHsShaderResource2(const HsShaderResource2& rsrc2) { m_vs.hs_regs.rsrc2 = rsrc2; }
-	void SetHsShaderChksum(uint32_t value) { m_vs.hs_regs.chksum = value; }
 	void SetGsShaderBase(uint64_t addr) { m_vs.gs_regs.data_addr = addr; }
 	void SetGsShaderResource1(const GsShaderResource1& rsrc1) { m_vs.gs_regs.rsrc1 = rsrc1; }
 	void SetGsShaderResource2(const GsShaderResource2& rsrc2) { m_vs.gs_regs.rsrc2 = rsrc2; }
-	void SetGsShaderChksum(uint32_t value) { m_vs.gs_regs.chksum = value; }
 
 	void SetPsShaderBase(uint64_t addr) { m_ps.ps_regs.data_addr = addr; }
 	void SetPsShaderResource1(const PsShaderResource1& rsrc1) { m_ps.ps_regs.rsrc1 = rsrc1; }
 	void SetPsShaderResource2(const PsShaderResource2& rsrc2) { m_ps.ps_regs.rsrc2 = rsrc2; }
-	void SetPsShaderChksum(uint32_t value) { m_ps.ps_regs.chksum = value; }
 
 	void SetCsShader(const CsStageRegisters& cs_regs) { m_cs.cs_regs = cs_regs; }
 	void SetCsWaveSize(uint8_t wave_size) { m_cs.cs_regs.wave_size = wave_size; }

src/graphics/guest_gpu/pm4.h

@@ -98,6 +98,10 @@ constexpr uint32_t DB_RENDER_CONTROL_DEPTH_CLEAR_ENABLE_SHIFT       = 0;
 constexpr uint32_t DB_RENDER_CONTROL_DEPTH_CLEAR_ENABLE_MASK        = 0x1;
 constexpr uint32_t DB_RENDER_CONTROL_STENCIL_CLEAR_ENABLE_SHIFT     = 1;
 constexpr uint32_t DB_RENDER_CONTROL_STENCIL_CLEAR_ENABLE_MASK      = 0x1;
+constexpr uint32_t DB_RENDER_CONTROL_COPY_DEPTH_TO_COLOR_SHIFT      = 2;
+constexpr uint32_t DB_RENDER_CONTROL_COPY_DEPTH_TO_COLOR_MASK       = 0x1;
+constexpr uint32_t DB_RENDER_CONTROL_COPY_STENCIL_TO_COLOR_SHIFT    = 3;
+constexpr uint32_t DB_RENDER_CONTROL_COPY_STENCIL_TO_COLOR_MASK     = 0x1;
 constexpr uint32_t DB_RENDER_CONTROL_RESUMMARIZE_ENABLE_SHIFT       = 4;
 constexpr uint32_t DB_RENDER_CONTROL_RESUMMARIZE_ENABLE_MASK        = 0x1;
 constexpr uint32_t DB_RENDER_CONTROL_STENCIL_COMPRESS_DISABLE_SHIFT = 5;
@@ -163,22 +167,6 @@ constexpr uint32_t PA_SC_SCREEN_SCISSOR_BR_BR_Y_MASK  = 0xFFFF;
 
 constexpr uint32_t DB_DFSM_CONTROL = 0xE;
 
-constexpr uint32_t DB_DEPTH_INFO                          = 0xF;
-constexpr uint32_t DB_DEPTH_INFO_ADDR5_SWIZZLE_MASK_SHIFT = 0;
-constexpr uint32_t DB_DEPTH_INFO_ADDR5_SWIZZLE_MASK_MASK  = 0xF;
-constexpr uint32_t DB_DEPTH_INFO_ARRAY_MODE_SHIFT         = 4;
-constexpr uint32_t DB_DEPTH_INFO_ARRAY_MODE_MASK          = 0xF;
-constexpr uint32_t DB_DEPTH_INFO_PIPE_CONFIG_SHIFT        = 8;
-constexpr uint32_t DB_DEPTH_INFO_PIPE_CONFIG_MASK         = 0x1F;
-constexpr uint32_t DB_DEPTH_INFO_BANK_WIDTH_SHIFT         = 13;
-constexpr uint32_t DB_DEPTH_INFO_BANK_WIDTH_MASK          = 0x3;
-constexpr uint32_t DB_DEPTH_INFO_BANK_HEIGHT_SHIFT        = 15;
-constexpr uint32_t DB_DEPTH_INFO_BANK_HEIGHT_MASK         = 0x3;
-constexpr uint32_t DB_DEPTH_INFO_MACRO_TILE_ASPECT_SHIFT  = 17;
-constexpr uint32_t DB_DEPTH_INFO_MACRO_TILE_ASPECT_MASK   = 0x3;
-constexpr uint32_t DB_DEPTH_INFO_NUM_BANKS_SHIFT          = 19;
-constexpr uint32_t DB_DEPTH_INFO_NUM_BANKS_MASK           = 0x3;
-
 constexpr uint32_t DB_Z_INFO                               = 0x10;
 constexpr uint32_t DB_Z_INFO_FORMAT_SHIFT                  = 0;
 constexpr uint32_t DB_Z_INFO_FORMAT_MASK                   = 0x3;
@@ -222,16 +210,6 @@ constexpr uint32_t DB_STENCIL_READ_BASE  = 0x13;
 constexpr uint32_t DB_Z_WRITE_BASE       = 0x14;
 constexpr uint32_t DB_STENCIL_WRITE_BASE = 0x15;
 
-constexpr uint32_t DB_DEPTH_SIZE                       = 0x16;
-constexpr uint32_t DB_DEPTH_SIZE_PITCH_TILE_MAX_SHIFT  = 0;
-constexpr uint32_t DB_DEPTH_SIZE_PITCH_TILE_MAX_MASK   = 0x7FF;
-constexpr uint32_t DB_DEPTH_SIZE_HEIGHT_TILE_MAX_SHIFT = 11;
-constexpr uint32_t DB_DEPTH_SIZE_HEIGHT_TILE_MAX_MASK  = 0x7FF;
-
-constexpr uint32_t DB_DEPTH_SLICE                      = 0x17;
-constexpr uint32_t DB_DEPTH_SLICE_SLICE_TILE_MAX_SHIFT = 0;
-constexpr uint32_t DB_DEPTH_SLICE_SLICE_TILE_MAX_MASK  = 0x3FFFFF;
-
 constexpr uint32_t DB_Z_READ_BASE_HI                  = 0x1A;
 constexpr uint32_t DB_STENCIL_READ_BASE_HI            = 0x1B;
 constexpr uint32_t DB_Z_WRITE_BASE_HI                 = 0x1C;
@@ -424,11 +402,6 @@ constexpr uint32_t DB_DEPTH_CONTROL_STENCILFUNC_SHIFT                        = 8
 constexpr uint32_t DB_DEPTH_CONTROL_STENCILFUNC_MASK                         = 0x7;
 constexpr uint32_t DB_DEPTH_CONTROL_STENCILFUNC_BF_SHIFT                     = 20;
 constexpr uint32_t DB_DEPTH_CONTROL_STENCILFUNC_BF_MASK                      = 0x7;
-constexpr uint32_t DB_DEPTH_CONTROL_ENABLE_COLOR_WRITES_ON_DEPTH_FAIL_SHIFT  = 30;
-constexpr uint32_t DB_DEPTH_CONTROL_ENABLE_COLOR_WRITES_ON_DEPTH_FAIL_MASK   = 0x1;
-constexpr uint32_t DB_DEPTH_CONTROL_DISABLE_COLOR_WRITES_ON_DEPTH_PASS_SHIFT = 31;
-constexpr uint32_t DB_DEPTH_CONTROL_DISABLE_COLOR_WRITES_ON_DEPTH_PASS_MASK  = 0x1;
-
 constexpr uint32_t DB_EQAA                                  = 0x201;
 constexpr uint32_t DB_EQAA_MAX_ANCHOR_SAMPLES_SHIFT         = 0;
 constexpr uint32_t DB_EQAA_MAX_ANCHOR_SAMPLES_MASK          = 0x7;
@@ -560,21 +533,9 @@ constexpr uint32_t VGT_DRAW_PAYLOAD_CNTL  = 0x2A6;
 constexpr uint32_t VGT_ESGS_RING_ITEMSIZE = 0x2AB;
 constexpr uint32_t VGT_REUSE_OFF          = 0x2AD;
 
-constexpr uint32_t DB_HTILE_SURFACE                               = 0x2AF;
-constexpr uint32_t DB_HTILE_SURFACE_LINEAR_SHIFT                  = 0;
-constexpr uint32_t DB_HTILE_SURFACE_LINEAR_MASK                   = 0x1;
-constexpr uint32_t DB_HTILE_SURFACE_FULL_CACHE_SHIFT              = 1;
-constexpr uint32_t DB_HTILE_SURFACE_FULL_CACHE_MASK               = 0x1;
-constexpr uint32_t DB_HTILE_SURFACE_HTILE_USES_PRELOAD_WIN_SHIFT  = 2;
-constexpr uint32_t DB_HTILE_SURFACE_HTILE_USES_PRELOAD_WIN_MASK   = 0x1;
-constexpr uint32_t DB_HTILE_SURFACE_PRELOAD_SHIFT                 = 3;
-constexpr uint32_t DB_HTILE_SURFACE_PRELOAD_MASK                  = 0x1;
-constexpr uint32_t DB_HTILE_SURFACE_PREFETCH_WIDTH_SHIFT          = 4;
-constexpr uint32_t DB_HTILE_SURFACE_PREFETCH_WIDTH_MASK           = 0x3F;
-constexpr uint32_t DB_HTILE_SURFACE_PREFETCH_HEIGHT_SHIFT         = 10;
-constexpr uint32_t DB_HTILE_SURFACE_PREFETCH_HEIGHT_MASK          = 0x3F;
-constexpr uint32_t DB_HTILE_SURFACE_DST_OUTSIDE_ZERO_TO_ONE_SHIFT = 16;
-constexpr uint32_t DB_HTILE_SURFACE_DST_OUTSIDE_ZERO_TO_ONE_MASK  = 0x1;
+constexpr uint32_t DB_SHADING_RATE_ENCODING                             = 0x2AF;
+constexpr uint32_t DB_SHADING_RATE_ENCODING_SHADING_RATE_ENCODING_SHIFT = 19;
+constexpr uint32_t DB_SHADING_RATE_ENCODING_SHADING_RATE_ENCODING_MASK  = 0x3;
 
 constexpr uint32_t DB_SRESULTS_COMPARE_STATE0                          = 0x2B0;
 constexpr uint32_t DB_SRESULTS_COMPARE_STATE1                          = 0x2B1;
@@ -621,7 +582,6 @@ constexpr uint32_t PA_SC_SHADER_CONTROL                  = 0x310;
 constexpr uint32_t PA_SC_BINNER_CNTL_0                   = 0x311;
 constexpr uint32_t PA_SC_BINNER_CNTL_1                   = 0x312;
 constexpr uint32_t PA_SC_CONSERVATIVE_RASTERIZATION_CNTL = 0x313;
-constexpr uint32_t PA_SC_NGG_MODE_CNTL                   = 0x314;
 constexpr uint32_t CB_COLOR0_BASE                        = 0x318;
 
 constexpr uint32_t CB_COLOR0_VIEW                   = 0x31B;
@@ -649,24 +609,13 @@ constexpr uint32_t CB_COLOR0_INFO_BLEND_BYPASS_SHIFT              = 16;
 constexpr uint32_t CB_COLOR0_INFO_BLEND_BYPASS_MASK               = 0x1;
 constexpr uint32_t CB_COLOR0_INFO_ROUND_MODE_SHIFT                = 18;
 constexpr uint32_t CB_COLOR0_INFO_ROUND_MODE_MASK                 = 0x1;
-constexpr uint32_t CB_COLOR0_INFO_CMASK_IS_LINEAR_SHIFT           = 19;
-constexpr uint32_t CB_COLOR0_INFO_CMASK_IS_LINEAR_MASK            = 0x1;
 constexpr uint32_t CB_COLOR0_INFO_FMASK_COMPRESSION_DISABLE_SHIFT = 26;
 constexpr uint32_t CB_COLOR0_INFO_FMASK_COMPRESSION_DISABLE_MASK  = 0x1;
 constexpr uint32_t CB_COLOR0_INFO_FMASK_COMPRESS_1FRAG_ONLY_SHIFT = 27;
 constexpr uint32_t CB_COLOR0_INFO_FMASK_COMPRESS_1FRAG_ONLY_MASK  = 0x1;
 constexpr uint32_t CB_COLOR0_INFO_DCC_ENABLE_SHIFT                = 28;
 constexpr uint32_t CB_COLOR0_INFO_DCC_ENABLE_MASK                 = 0x1;
-constexpr uint32_t CB_COLOR0_INFO_CMASK_ADDR_TYPE_SHIFT           = 29;
-constexpr uint32_t CB_COLOR0_INFO_CMASK_ADDR_TYPE_MASK            = 0x3;
-constexpr uint32_t CB_COLOR0_INFO_ALT_TILE_MODE_SHIFT             = 31;
-constexpr uint32_t CB_COLOR0_INFO_ALT_TILE_MODE_MASK              = 0x1;
-
 constexpr uint32_t CB_COLOR0_ATTRIB                             = 0x31D;
-constexpr uint32_t CB_COLOR0_ATTRIB_TILE_MODE_INDEX_SHIFT       = 0;
-constexpr uint32_t CB_COLOR0_ATTRIB_TILE_MODE_INDEX_MASK        = 0x1F;
-constexpr uint32_t CB_COLOR0_ATTRIB_FMASK_TILE_MODE_INDEX_SHIFT = 5;
-constexpr uint32_t CB_COLOR0_ATTRIB_FMASK_TILE_MODE_INDEX_MASK  = 0x1F;
 constexpr uint32_t CB_COLOR0_ATTRIB_NUM_SAMPLES_SHIFT           = 12;
 constexpr uint32_t CB_COLOR0_ATTRIB_NUM_SAMPLES_MASK            = 0x7;
 constexpr uint32_t CB_COLOR0_ATTRIB_NUM_FRAGMENTS_SHIFT         = 15;
@@ -681,8 +630,6 @@ constexpr uint32_t CB_COLOR0_DCC_CONTROL_KEY_CLEAR_ENABLE_SHIFT
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_KEY_CLEAR_ENABLE_MASK                  = 0x1;
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_MAX_UNCOMPRESSED_BLOCK_SIZE_SHIFT      = 2;
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_MAX_UNCOMPRESSED_BLOCK_SIZE_MASK       = 0x3;
-constexpr uint32_t CB_COLOR0_DCC_CONTROL_MIN_COMPRESSED_BLOCK_SIZE_SHIFT        = 4;
-constexpr uint32_t CB_COLOR0_DCC_CONTROL_MIN_COMPRESSED_BLOCK_SIZE_MASK         = 0x1;
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_MAX_COMPRESSED_BLOCK_SIZE_SHIFT        = 5;
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_MAX_COMPRESSED_BLOCK_SIZE_MASK         = 0x3;
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_COLOR_TRANSFORM_SHIFT                  = 7;
@@ -739,31 +686,25 @@ constexpr uint32_t CB_COLOR0_ATTRIB3_CMASK_PIPE_ALIGNED_SHIFT = 26;
 constexpr uint32_t CB_COLOR0_ATTRIB3_CMASK_PIPE_ALIGNED_MASK  = 0x1;
 constexpr uint32_t CB_COLOR0_ATTRIB3_DCC_PIPE_ALIGNED_SHIFT   = 30;
 constexpr uint32_t CB_COLOR0_ATTRIB3_DCC_PIPE_ALIGNED_MASK    = 0x1;
+constexpr uint32_t CB_COLOR0_ATTRIB3_WRITE_VRS_RATE_HINT_TO_CMASK_SHIFT = 31;
+constexpr uint32_t CB_COLOR0_ATTRIB3_WRITE_VRS_RATE_HINT_TO_CMASK_MASK  = 0x1;
 
 constexpr uint32_t CB_COLOR7_ATTRIB3 = 0x3BF;
 
-/* Fake codes. Don't exist on real HW */
-
-constexpr uint32_t CX_NOP = 0x800003FF;
+constexpr uint32_t CX_NOP = 0xDB;
 
 constexpr uint32_t CX_NUM = 0x3FF + 1;
 
 /* Shader registers */
 
-constexpr uint32_t SPI_SHADER_TBA_LO_PS     = 0x0;
-constexpr uint32_t SPI_SHADER_PGM_RSRC4_PS  = 0x1;
-constexpr uint32_t SPI_SHADER_TMA_LO_PS     = 0x2;
-constexpr uint32_t SPI_SHADER_TMA_HI_PS     = 0x3;
-constexpr uint32_t SPI_SHADER_PGM_CHKSUM_PS = 0x6;
-constexpr uint32_t SPI_SHADER_PGM_RSRC3_PS  = 0x7;
-constexpr uint32_t SPI_SHADER_PGM_LO_PS     = 0x8;
-constexpr uint32_t SPI_SHADER_PGM_HI_PS     = 0x9;
+constexpr uint32_t SPI_SHADER_PACE_ID_PS            = 0x6;
+constexpr uint32_t SPI_GRAPHICS_SHADER_CONTROL_PS    = 0x7;
+constexpr uint32_t SPI_SHADER_PGM_LO_PS              = 0x8;
+constexpr uint32_t SPI_SHADER_PGM_HI_PS              = 0x9;
 
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS                        = 0xA;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_VGPRS_SHIFT            = 0;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_VGPRS_MASK             = 0x3F;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_SGPRS_SHIFT            = 6;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_SGPRS_MASK             = 0xF;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_PRIORITY_SHIFT         = 10;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_PRIORITY_MASK          = 0x3;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_FLOAT_MODE_SHIFT       = 12;
@@ -800,14 +741,11 @@ constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_SHARED_VGPR_CNT_MASK         = 0xF;
 constexpr uint32_t SPI_SHADER_USER_DATA_PS_0  = 0xC;
 constexpr uint32_t SPI_SHADER_USER_DATA_PS_15 = 0x1B;
 constexpr uint32_t SPI_SHADER_USER_DATA_PS_31 = 0x2B;
-constexpr uint32_t SPI_SHADER_REQ_CTRL_PS     = 0x30;
 constexpr uint32_t SPI_SHADER_USER_ACCUM_PS_0 = 0x32;
 constexpr uint32_t SPI_SHADER_USER_ACCUM_PS_3 = 0x35;
 
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_VGPRS_SHIFT           = 0;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_VGPRS_MASK            = 0x3F;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_SGPRS_SHIFT           = 6;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_SGPRS_MASK            = 0xF;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_PRIORITY_SHIFT        = 10;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_PRIORITY_MASK         = 0x3;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_FLOAT_MODE_SHIFT      = 12;
@@ -838,19 +776,20 @@ constexpr uint32_t SPI_SHADER_PGM_RSRC2_VS_USER_SGPR_MSB_MASK    = 0x1;
 constexpr uint32_t SPI_SHADER_PGM_RSRC2_VS_SHARED_VGPR_CNT_SHIFT = 28;
 constexpr uint32_t SPI_SHADER_PGM_RSRC2_VS_SHARED_VGPR_CNT_MASK  = 0xF;
 
-constexpr uint32_t SPI_SHADER_PGM_CHKSUM_GS        = 0x80;
+constexpr uint32_t SPI_SHADER_PACE_ID_GS           = 0x80;
+// AGC compiler records use this numeric tag as opaque metadata. It is not a shader checksum
+// register and intentionally has no checksum-state consumer in HardwareContext.
+constexpr uint32_t SPI_SHADER_PGM_CHKSUM_GS        = SPI_SHADER_PACE_ID_GS;
 constexpr uint32_t SPI_SHADER_PGM_RSRC4_GS         = 0x81;
 constexpr uint32_t SPI_SHADER_USER_DATA_ADDR_LO_GS = 0x82;
 constexpr uint32_t SPI_SHADER_USER_DATA_ADDR_HI_GS = 0x83;
-constexpr uint32_t SPI_SHADER_PGM_RSRC3_GS         = 0x87;
+constexpr uint32_t SPI_GRAPHICS_SHADER_CONTROL_GS  = 0x87;
 constexpr uint32_t SPI_SHADER_PGM_LO_GS            = 0x88;
 constexpr uint32_t SPI_SHADER_PGM_HI_GS            = 0x89;
 
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS                        = 0x8A;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_VGPRS_SHIFT            = 0;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_VGPRS_MASK             = 0x3F;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_SGPRS_SHIFT            = 6;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_SGPRS_MASK             = 0xF;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_PRIORITY_SHIFT         = 10;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_PRIORITY_MASK          = 0x3;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_FLOAT_MODE_SHIFT       = 12;
@@ -891,22 +830,16 @@ constexpr uint32_t SPI_SHADER_PGM_RSRC2_GS_SHARED_VGPR_CNT_MASK   = 0xF;
 constexpr uint32_t SPI_SHADER_USER_DATA_GS_0                      = 0x8C;
 constexpr uint32_t SPI_SHADER_USER_DATA_GS_15                     = 0x9B;
 constexpr uint32_t SPI_SHADER_USER_DATA_GS_31                     = 0xAB;
-constexpr uint32_t SPI_SHADER_REQ_CTRL_ESGS                       = 0xB0;
 constexpr uint32_t SPI_SHADER_USER_ACCUM_ESGS_0                   = 0xB2;
 constexpr uint32_t SPI_SHADER_USER_ACCUM_ESGS_3                   = 0xB5;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_ES_GS                     = 0xBC;
 constexpr uint32_t SPI_SHADER_PGM_LO_ES                           = 0xC8;
 constexpr uint32_t SPI_SHADER_PGM_HI_ES                           = 0xC9;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_ES                        = 0xCA;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_ES                        = 0xCB;
-constexpr uint32_t SPI_SHADER_USER_DATA_ES_0                      = 0xCC;
-constexpr uint32_t SPI_SHADER_USER_DATA_ES_15                     = 0xDB;
-constexpr uint32_t SPI_SHADER_USER_DATA_ES_31                     = 0xEB;
+// Private AGC registers used by shader binaries submitted through SET_SH_REG_INDIRECT.
 constexpr uint32_t SPI_SHADER_PGM_CHKSUM_HS                       = 0x100;
 constexpr uint32_t SPI_SHADER_PGM_RSRC4_HS                        = 0x101;
 constexpr uint32_t SPI_SHADER_USER_DATA_ADDR_LO_HS                = 0x102;
 constexpr uint32_t SPI_SHADER_USER_DATA_ADDR_HI_HS                = 0x103;
-constexpr uint32_t SPI_SHADER_PGM_RSRC3_HS                        = 0x107;
+constexpr uint32_t SPI_GRAPHICS_SHADER_CONTROL_HS                 = 0x107;
 constexpr uint32_t SPI_SHADER_PGM_LO_HS                           = 0x108;
 constexpr uint32_t SPI_SHADER_PGM_HI_HS                           = 0x109;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS                        = 0x10A;
@@ -944,13 +877,10 @@ constexpr uint32_t SPI_SHADER_PGM_RSRC2_HS_SHARED_VGPR_CNT_MASK   = 0xF;
 constexpr uint32_t SPI_SHADER_USER_DATA_HS_0                      = 0x10C;
 constexpr uint32_t SPI_SHADER_USER_DATA_HS_15                     = 0x11B;
 constexpr uint32_t SPI_SHADER_USER_DATA_HS_31                     = 0x12B;
-constexpr uint32_t SPI_SHADER_REQ_CTRL_LSHS                       = 0x130;
 constexpr uint32_t SPI_SHADER_USER_ACCUM_LSHS_0                   = 0x132;
 constexpr uint32_t SPI_SHADER_USER_ACCUM_LSHS_3                   = 0x135;
 constexpr uint32_t SPI_SHADER_PGM_LO_LS                           = 0x148;
 constexpr uint32_t SPI_SHADER_PGM_HI_LS                           = 0x149;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_LS                        = 0x14A;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_LS                        = 0x14B;
 constexpr uint32_t COMPUTE_START_X                                = 0x204;
 constexpr uint32_t COMPUTE_START_Y                                = 0x205;
 constexpr uint32_t COMPUTE_START_Z                                = 0x206;
@@ -959,28 +889,26 @@ constexpr uint32_t COMPUTE_NUM_THREAD_Y                           = 0x208;
 constexpr uint32_t COMPUTE_NUM_THREAD_Z                           = 0x209;
 constexpr uint32_t COMPUTE_PGM_LO                                 = 0x20C;
 constexpr uint32_t COMPUTE_PGM_HI                                 = 0x20D;
-constexpr uint32_t COMPUTE_TBA_LO                                 = 0x20E;
-constexpr uint32_t COMPUTE_TBA_HI                                 = 0x20F;
-constexpr uint32_t COMPUTE_TMA_LO                                 = 0x210;
-constexpr uint32_t COMPUTE_TMA_HI                                 = 0x211;
 
 constexpr uint32_t COMPUTE_PGM_RSRC1                  = 0x212;
 constexpr uint32_t COMPUTE_PGM_RSRC1_VGPRS_SHIFT      = 0;
 constexpr uint32_t COMPUTE_PGM_RSRC1_VGPRS_MASK       = 0x3F;
-constexpr uint32_t COMPUTE_PGM_RSRC1_SGPRS_SHIFT      = 6;
-constexpr uint32_t COMPUTE_PGM_RSRC1_SGPRS_MASK       = 0xF;
+constexpr uint32_t COMPUTE_PGM_RSRC1_PRIORITY_SHIFT   = 10;
+constexpr uint32_t COMPUTE_PGM_RSRC1_PRIORITY_MASK    = 0x3;
 constexpr uint32_t COMPUTE_PGM_RSRC1_FLOAT_MODE_SHIFT = 12;
 constexpr uint32_t COMPUTE_PGM_RSRC1_FLOAT_MODE_MASK  = 0xFF;
 constexpr uint32_t COMPUTE_PGM_RSRC1_DX10_CLAMP_SHIFT = 21;
 constexpr uint32_t COMPUTE_PGM_RSRC1_DX10_CLAMP_MASK  = 0x1;
+constexpr uint32_t COMPUTE_PGM_RSRC1_DEBUG_MODE_SHIFT = 22;
+constexpr uint32_t COMPUTE_PGM_RSRC1_DEBUG_MODE_MASK  = 0x1;
 constexpr uint32_t COMPUTE_PGM_RSRC1_IEEE_MODE_SHIFT  = 23;
 constexpr uint32_t COMPUTE_PGM_RSRC1_IEEE_MODE_MASK   = 0x1;
-constexpr uint32_t COMPUTE_PGM_RSRC1_BULKY_SHIFT      = 24;
-constexpr uint32_t COMPUTE_PGM_RSRC1_BULKY_MASK       = 0x1;
 constexpr uint32_t COMPUTE_PGM_RSRC1_FP16_OVFL_SHIFT  = 26;
 constexpr uint32_t COMPUTE_PGM_RSRC1_FP16_OVFL_MASK   = 0x1;
 constexpr uint32_t COMPUTE_PGM_RSRC1_WGP_MODE_SHIFT   = 29;
 constexpr uint32_t COMPUTE_PGM_RSRC1_WGP_MODE_MASK    = 0x1;
+constexpr uint32_t COMPUTE_PGM_RSRC1_FWD_PROGRESS_SHIFT = 31;
+constexpr uint32_t COMPUTE_PGM_RSRC1_FWD_PROGRESS_MASK  = 0x1;
 
 constexpr uint32_t COMPUTE_DISPATCH_INITIATOR_CS_W32_EN_SHIFT = 15;
 constexpr uint32_t COMPUTE_DISPATCH_INITIATOR_CS_W32_EN_MASK  = 0x1;
@@ -1011,21 +939,16 @@ constexpr uint32_t COMPUTE_PGM_RSRC2_LDS_SIZE_SHIFT       = 15;
 constexpr uint32_t COMPUTE_PGM_RSRC2_LDS_SIZE_MASK        = 0x1FF;
 
 constexpr uint32_t COMPUTE_RESOURCE_LIMITS    = 0x215;
-constexpr uint32_t COMPUTE_DESTINATION_EN_SE0 = 0x216;
-constexpr uint32_t COMPUTE_DESTINATION_EN_SE1 = 0x217;
 constexpr uint32_t COMPUTE_TMPRING_SIZE       = 0x218;
-constexpr uint32_t COMPUTE_DESTINATION_EN_SE2 = 0x219;
-constexpr uint32_t COMPUTE_DESTINATION_EN_SE3 = 0x21A;
 constexpr uint32_t COMPUTE_USER_ACCUM_0       = 0x224;
 constexpr uint32_t COMPUTE_PGM_RSRC3          = 0x228;
-constexpr uint32_t COMPUTE_SHADER_CHKSUM      = 0x22A;
+constexpr uint32_t COMPUTE_PGM_RSRC3_SHARED_VGPRS_SHIFT = 0;
+constexpr uint32_t COMPUTE_PGM_RSRC3_SHARED_VGPRS_MASK  = 0xF;
+constexpr uint32_t COMPUTE_PACE_ID             = 0x22A;
 constexpr uint32_t COMPUTE_USER_DATA_0        = 0x240;
 constexpr uint32_t COMPUTE_USER_DATA_15       = 0x24F;
-constexpr uint32_t COMPUTE_DISPATCH_TUNNEL    = 0x27D;
 
-/* Fake codes. Don't exist on real HW */
-
-constexpr uint32_t SH_NOP = 0x800002FF;
+constexpr uint32_t SH_NOP = 0x280;
 
 constexpr uint32_t SH_NUM = 0x2FF + 1;
 
@@ -1042,6 +965,7 @@ constexpr uint32_t GE_INDX_OFFSET            = 0x24A;
 constexpr uint32_t GE_MULTI_PRIM_IB_RESET_EN = 0x24B;
 constexpr uint32_t VGT_HS_OFFCHIP_PARAM      = 0x24F;
 constexpr uint32_t VGT_TF_MEMORY_BASE        = 0x250;
+constexpr uint32_t IA_MULTI_VGT_PARAM        = 0x258;
 
 constexpr uint32_t GE_CNTL                     = 0x25B;
 constexpr uint32_t GE_CNTL_PRIM_GRP_SIZE_SHIFT = 0;
@@ -1053,7 +977,7 @@ constexpr uint32_t GE_USER_VGPR1  = 0x25C;
 constexpr uint32_t GE_USER_VGPR2  = 0x25D;
 constexpr uint32_t GE_USER_VGPR3  = 0x25E;
 constexpr uint32_t GE_STEREO_CNTL = 0x25F;
-constexpr uint32_t GE_PC_ALLOC    = 0x260;
+constexpr uint32_t UC_PARAMETER_OVERSUBSCRIPTION = 0x260;
 
 constexpr uint32_t GE_USER_VGPR_EN                     = 0x262;
 constexpr uint32_t GE_USER_VGPR_EN_EN_USER_VGPR1_SHIFT = 0;
@@ -1063,28 +987,15 @@ constexpr uint32_t GE_USER_VGPR_EN_EN_USER_VGPR2_MASK  = 0x1;
 constexpr uint32_t GE_USER_VGPR_EN_EN_USER_VGPR3_SHIFT = 2;
 constexpr uint32_t GE_USER_VGPR_EN_EN_USER_VGPR3_MASK  = 0x1;
 
-constexpr uint32_t IA_MULTI_VGT_PARAM       = 0x258;
 constexpr uint32_t TA_CS_BC_BASE_ADDR       = 0x380;
 constexpr uint32_t TA_CS_BC_BASE_ADDR_HI    = 0x381;
-constexpr uint32_t TEXTURE_GRADIENT_FACTORS = 0x382;
+constexpr uint32_t TEXTURE_GRADIENT_FACTORS = 0x383;
+constexpr uint32_t TEXTURE_GRADIENT_CONTROL = 0x384;
 constexpr uint32_t GDS_OA_CNTL              = 0x41D;
 constexpr uint32_t GDS_OA_COUNTER           = 0x41E;
 constexpr uint32_t GDS_OA_ADDRESS           = 0x41F;
 
-/* Fake codes. Don't exist on real HW */
-
-constexpr uint32_t FSR_EXTEND_SUBPIXEL_ROUNDING = 0x80003FF4;
-constexpr uint32_t FSR_ALPHA_VALUE0             = 0x80003FF5;
-constexpr uint32_t FSR_ALPHA_VALUE1             = 0x80003FF6;
-constexpr uint32_t FSR_CONTROL_POINT0           = 0x80003FF7;
-constexpr uint32_t FSR_CONTROL_POINT1           = 0x80003FF8;
-constexpr uint32_t FSR_CONTROL_POINT2           = 0x80003FF9;
-constexpr uint32_t FSR_CONTROL_POINT3           = 0x80003FFA;
-constexpr uint32_t FSR_WINDOW0                  = 0x80003FFB;
-constexpr uint32_t FSR_WINDOW1                  = 0x80003FFC;
-constexpr uint32_t TEXTURE_GRADIENT_CONTROL     = 0x80003FFD;
-constexpr uint32_t MEMORY_MAPPING_MASK          = 0x80003FFE;
-constexpr uint32_t UC_NOP                       = 0x80003FFF;
+constexpr uint32_t UC_NOP = 0xA2;
 
 constexpr uint32_t UC_NUM = 0x3FFF + 1;
 

src/graphics/host_gpu/memoryTracker.cpp

@@ -18,8 +18,8 @@ MemoryTracker::~MemoryTracker() = default;
 #if KYTY_BUILD == KYTY_BUILD_DEBUG
 void MemoryTracker::ValidateGpuDirtyPages(const RangeSet& dirty, uint64_t vaddr, uint64_t size,
                                           const char* operation) const noexcept {
-	if (vaddr == 0 || size == 0 || size > UINT64_MAX - vaddr ||
-	    (vaddr & (TRACKER_PAGE_SIZE - 1)) != 0 || (size & (TRACKER_PAGE_SIZE - 1)) != 0) {
+	if (!GuestRange {vaddr, size}.Valid() || (vaddr & (TRACKER_PAGE_SIZE - 1)) != 0 ||
+	    (size & (TRACKER_PAGE_SIZE - 1)) != 0) {
 		EXIT("MemoryTracker: invalid dirty-page validation range\n");
 	}
 	for (auto page = vaddr; page < vaddr + size; page += TRACKER_PAGE_SIZE) {
@@ -37,9 +37,6 @@ void MemoryTracker::ValidateGpuDirtyPages(const RangeSet& dirty, uint64_t vaddr,
 void MemoryTracker::ValidateGpuDirtyOwnership(const RangeSet& dirty, uint64_t vaddr, uint64_t size,
                                               const char* operation) {
 	ValidateRange(vaddr, size);
-	if (vaddr + size > UINT64_MAX - (TRACKER_PAGE_SIZE - 1)) {
-		EXIT("MemoryTracker: dirty ownership range alignment overflow\n");
-	}
 	const auto begin = vaddr & ~(TRACKER_PAGE_SIZE - 1);
 	const auto end   = (vaddr + size + TRACKER_PAGE_SIZE - 1) & ~(TRACKER_PAGE_SIZE - 1);
 	for (auto page = begin; page < end; page += TRACKER_PAGE_SIZE) {
@@ -56,8 +53,7 @@ void MemoryTracker::ValidateGpuDirtyOwnership(const RangeSet& dirty, uint64_t va
 #endif
 
 void MemoryTracker::ValidateRange(uint64_t vaddr, uint64_t size) {
-	if (vaddr == 0 || size == 0 || vaddr >= TRACKER_ADDRESS_SIZE ||
-	    size > TRACKER_ADDRESS_SIZE - vaddr) {
+	if (!GuestRange {vaddr, size}.Valid()) {
 		EXIT("invalid memory tracker range\n");
 	}
 }

src/graphics/host_gpu/pageManager.cpp

@@ -110,7 +110,7 @@ private:
 };
 
 void ValidateRange(uint64_t vaddr, uint64_t size) {
-	if (vaddr == 0 || size == 0 || vaddr >= ADDRESS_SIZE || size > ADDRESS_SIZE - vaddr) {
+	if (!GuestRange {vaddr, size}.Valid()) {
 		Fatal("invalid range vaddr=0x%016" PRIx64 ", size=0x%016" PRIx64, vaddr, size);
 	}
 }

src/graphics/host_gpu/regionDefinitions.h

@@ -4,6 +4,8 @@
 #include "common/bitArray.h"
 #include "common/common.h"
 
+#include <compare>
+
 namespace Libs::Graphics {
 
 constexpr uint64_t TRACKER_PAGE_SIZE    = 4ull * 1024ull;
@@ -11,6 +13,20 @@ constexpr uint64_t TRACKER_REGION_SIZE  = 4ull * 1024ull * 1024ull;
 constexpr uint64_t TRACKER_ADDRESS_SIZE = 1ull << 40u;
 constexpr size_t   TRACKER_REGION_PAGES = TRACKER_REGION_SIZE / TRACKER_PAGE_SIZE;
 
+struct GuestRange {
+	uint64_t address = 0;
+	uint64_t size    = 0;
+
+	[[nodiscard]] constexpr bool Empty() const noexcept { return address == 0 && size == 0; }
+	[[nodiscard]] constexpr bool Valid() const noexcept {
+		return address != 0 && size != 0 && address < TRACKER_ADDRESS_SIZE &&
+		       size <= TRACKER_ADDRESS_SIZE - address;
+	}
+	[[nodiscard]] constexpr bool     ValidOrEmpty() const noexcept { return Empty() || Valid(); }
+	[[nodiscard]] constexpr uint64_t End() const noexcept { return address + size; }
+	auto                             operator<=>(const GuestRange&) const = default;
+};
+
 enum class DirtySource { Cpu, Gpu };
 using RegionBits = Common::BitArray<TRACKER_REGION_PAGES>;
 static_assert(sizeof(RegionBits) == TRACKER_REGION_PAGES / 8);

src/graphics/host_gpu/renderer/cache/bufferCache.cpp

@@ -241,8 +241,7 @@ BufferCache::~BufferCache() {
 }
 
 void BufferCache::InvalidateMemory(uint64_t vaddr, uint64_t size) {
-	if (vaddr == 0 || size == 0 || vaddr >= TRACKER_ADDRESS_SIZE ||
-	    size > TRACKER_ADDRESS_SIZE - vaddr) {
+	if (!GuestRange {vaddr, size}.Valid()) {
 		EXIT("BufferCache: invalid memory-invalidation range\n");
 	}
 	m_memory_tracker.InvalidateRegion(vaddr, size,
@@ -314,7 +313,7 @@ BufferId BufferCache::FindBuffer(uint64_t vaddr, uint64_t size) {
 	if (vaddr == 0) {
 		return NULL_BUFFER_ID;
 	}
-	if (size == 0 || vaddr >= TRACKER_ADDRESS_SIZE || size > TRACKER_ADDRESS_SIZE - vaddr) {
+	if (!GuestRange {vaddr, size}.Valid()) {
 		EXIT("BufferCache: invalid buffer discovery request\n");
 	}
 	const auto* owner = m_page_table.Find(vaddr >> PageTable::kPageBits);
@@ -443,8 +442,7 @@ std::pair<Buffer*, uint64_t> BufferCache::ObtainBuffer(uint64_t vaddr, uint64_t
                                                        bool is_written, bool is_texel_buffer,
                                                        BufferId id) {
 	auto& command = m_scheduler.Current();
-	if (command.IsInvalid() || vaddr == 0 || size == 0 ||
-	    vaddr >= TRACKER_ADDRESS_SIZE || size > TRACKER_ADDRESS_SIZE - vaddr) {
+	if (command.IsInvalid() || !GuestRange {vaddr, size}.Valid()) {
 		EXIT("BufferCache: buffer request requires a recording command buffer\n");
 	}
 
@@ -474,8 +472,7 @@ std::pair<Buffer*, uint64_t> BufferCache::ObtainBuffer(uint64_t vaddr, uint64_t
 }
 
 std::pair<Buffer*, uint64_t> BufferCache::ObtainBufferForImage(uint64_t vaddr, uint64_t size) {
-	if (vaddr == 0 || size == 0 || vaddr >= TRACKER_ADDRESS_SIZE ||
-	    size > TRACKER_ADDRESS_SIZE - vaddr) {
+	if (!GuestRange {vaddr, size}.Valid()) {
 		EXIT("BufferCache: invalid image source\n");
 	}
 	auto find_owner = [&]() -> Buffer* {
@@ -661,8 +658,7 @@ void BufferCache::CopyBuffer(uint64_t dst_vaddr, uint64_t src_vaddr, uint64_t si
 }
 
 bool BufferCache::IsRegionRegistered(uint64_t vaddr, uint64_t size) {
-	if (vaddr == 0 || size == 0 || vaddr >= TRACKER_ADDRESS_SIZE ||
-	    size > TRACKER_ADDRESS_SIZE - vaddr) {
+	if (!GuestRange {vaddr, size}.Valid()) {
 		EXIT("BufferCache: invalid registered-region query\n");
 	}
 	// Cached buffers are ordered and non-overlapping. The last buffer beginning before the query

src/graphics/host_gpu/renderer/cache/gpuResourceManager.cpp

@@ -35,8 +35,7 @@ bool GpuResourceManager::InvalidateMemory(uint64_t vaddr, uint64_t size) {
 }
 
 bool GpuResourceManager::IsMapped(uint64_t vaddr, uint64_t size) const noexcept {
-	if (vaddr == 0 || size == 0 || vaddr >= TRACKER_ADDRESS_SIZE ||
-	    size > TRACKER_ADDRESS_SIZE - vaddr) {
+	if (!GuestRange {vaddr, size}.Valid()) {
 		return false;
 	}
 	std::shared_lock lock(m_mapped_ranges_mutex);

src/graphics/host_gpu/renderer/colorRenderTarget.cpp

@@ -204,13 +204,11 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, CommandBuffer&
 	if (texture_tile &&
 	    (!TileGetTextureBlockLayout(transfer_format, rt.attrib3.tile_mode, volume,
 	                                texture_tile_layout) ||
-	     rt.pitch.pitch_div8_minus1 != 0 ||
 	     (rt.base.addr & (texture_tile_layout.block.block_size - 1u)) != 0 ||
 	     rt.info.fmask_compression_enable || rt.info.fmask_data_compression_disable ||
 	     rt.info.fmask_one_frag_mode || rt.info.cmask_fast_clear_enable ||
-	     rt.info.dcc_compression_enable || rt.info.cmask_is_linear != 0 ||
-	     rt.info.cmask_addr_type != 0 || rt.info.alt_tile_mode || rt.cmask.addr != 0 ||
-	     rt.fmask.addr != 0 || rt.dcc_addr.addr != 0 || rt.dcc.data_write_on_dcc_clear_to_reg)) {
+	     rt.info.dcc_compression_enable || rt.cmask.addr != 0 || rt.fmask.addr != 0 ||
+	     rt.dcc_addr.addr != 0 || rt.dcc.data_write_on_dcc_clear_to_reg)) {
 		EXIT("unsupported texture-tiled render target: addr=0x%016" PRIx64 " tile=%u"
 		     " dimension=%u depth=%u levels=%u layer=%u/%u samples=%u fragments=%u bpe=%u"
 		     " cmask=0x%016" PRIx64 " fmask=0x%016" PRIx64 " dcc=0x%016" PRIx64 "\n",
@@ -226,9 +224,7 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, CommandBuffer&
 		     " layer=%u/%u\n",
 		     rt.attrib3.dimension, rt.attrib3.depth, view.base_layer, view.image_layers);
 	}
-	if (rt.pitch.pitch_div8_minus1 != 0) {
-		pitch = (rt.pitch.pitch_div8_minus1 + 1u) << 3u;
-	} else if (tile) {
+	if (tile) {
 		if (volume) {
 			pitch = TileGetTexturePitch(transfer_format, width, rt.attrib3.tile_mode);
 		} else if (texture_tile) {
@@ -294,12 +290,6 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, CommandBuffer&
 		mip_sizes[0]  = {static_cast<uint32_t>(size), 0, 0, 0, 0, 0};
 		mip_padded[0] = {pitch, height};
 	}
-	if (rt.slice.slice_div64_minus1 != 0 &&
-	    (static_cast<uint64_t>(rt.slice.slice_div64_minus1) + 1u) * 64u != size) {
-		EXIT("render-target slice span mismatch: encoded=0x%016" PRIx64 " derived=0x%016" PRIx64
-		     "\n",
-		     (static_cast<uint64_t>(rt.slice.slice_div64_minus1) + 1u) * 64u, size);
-	}
 	if (size == 0 || (!volume && size > UINT64_MAX / view.image_layers)) {
 		EXIT("render-target memory footprint is invalid\n");
 	}
@@ -309,7 +299,7 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, CommandBuffer&
 	if (backing_size == 0) {
 		EXIT("render-target backing is empty\n");
 	}
-	if (backing_size > TRACKER_ADDRESS_SIZE - rt.base.addr) {
+	if (!GuestRange {rt.base.addr, backing_size}.Valid()) {
 		EXIT("render-target backing range is invalid\n");
 	}
 

src/graphics/host_gpu/renderer/debug.cpp

@@ -97,17 +97,11 @@ void sh_print(const char* func, const HW::Shader& /*uc*/) {
 
 std::vector<std::string> rt_print(const char* func, const HW::RenderTarget& rt) {
 	std::vector<std::string> dst;
-	dst.reserve(53);
+	dst.reserve(40);
 
 	dst.push_back(fmt::format("{}\n", func));
 
 	dst.push_back(fmt::format("\t base.addr                       = 0x{:016x}\n", rt.base.addr));
-	dst.push_back(
-	    fmt::format("\t pitch.pitch_div8_minus1         = 0x{:08x}\n", rt.pitch.pitch_div8_minus1));
-	dst.push_back(fmt::format("\t pitch.fmask_pitch_div8_minus1   = 0x{:08x}\n",
-	                          rt.pitch.fmask_pitch_div8_minus1));
-	dst.push_back(fmt::format("\t slice.slice_div64_minus1        = 0x{:08x}\n",
-	                          rt.slice.slice_div64_minus1));
 	dst.push_back(fmt::format("\t view.base_array_slice_index     = 0x{:08x}\n",
 	                          rt.view.base_array_slice_index));
 	dst.push_back(fmt::format("\t view.last_array_slice_index     = 0x{:08x}\n",
@@ -128,12 +122,6 @@ std::vector<std::string> rt_print(const char* func, const HW::RenderTarget& rt)
 	                          rt.info.cmask_fast_clear_enable ? "true" : "false"));
 	dst.push_back(fmt::format("\t info.dcc_compression_enable     = {}\n",
 	                          rt.info.dcc_compression_enable ? "true" : "false"));
-	dst.push_back(
-	    fmt::format("\t info.cmask_is_linear            = 0x{:08x}\n", rt.info.cmask_is_linear));
-	dst.push_back(
-	    fmt::format("\t info.cmask_addr_type            = 0x{:08x}\n", rt.info.cmask_addr_type));
-	dst.push_back(fmt::format("\t info.alt_tile_mode              = {}\n",
-	                          rt.info.alt_tile_mode ? "true" : "false"));
 	dst.push_back(fmt::format("\t info.format                     = 0x{:08x}\n",
 	                          static_cast<uint32_t>(rt.info.format)));
 	dst.push_back(fmt::format("\t info.channel_type               = 0x{:08x}\n",
@@ -148,10 +136,6 @@ std::vector<std::string> rt_print(const char* func, const HW::RenderTarget& rt)
 	                          rt.info.round_mode ? "true" : "false"));
 	dst.push_back(fmt::format("\t attrib.force_dest_alpha_to_one  = {}\n",
 	                          rt.attrib.force_dest_alpha_to_one ? "true" : "false"));
-	dst.push_back(fmt::format("\t attrib.tile_mode                = 0x{:08x}\n",
-	                          static_cast<uint32_t>(rt.attrib.tile_mode)));
-	dst.push_back(fmt::format("\t attrib.fmask_tile_mode          = 0x{:08x}\n",
-	                          static_cast<uint32_t>(rt.attrib.fmask_tile_mode)));
 	dst.push_back(
 	    fmt::format("\t attrib.num_samples              = 0x{:08x}\n", rt.attrib.num_samples));
 	dst.push_back(
@@ -166,42 +150,32 @@ std::vector<std::string> rt_print(const char* func, const HW::RenderTarget& rt)
 	                          static_cast<uint32_t>(rt.attrib3.tile_mode)));
 	dst.push_back(
 	    fmt::format("\t attrib3.dimension               = 0x{:08x}\n", rt.attrib3.dimension));
-	dst.push_back(fmt::format("\t attrib3.cmask_pipe_aligned      = {}\n",
-	                          rt.attrib3.cmask_pipe_aligned ? "true" : "false"));
-	dst.push_back(fmt::format("\t attrib3.dcc_pipe_aligned        = {}\n",
-	                          rt.attrib3.dcc_pipe_aligned ? "true" : "false"));
+	dst.push_back(fmt::format("\t attrib3.metadata_pipe_aligned   = {}\n",
+	                          rt.attrib3.metadata_pipe_aligned ? "true" : "false"));
+	dst.push_back(fmt::format("\t attrib3.write_vrs_rate_hint_to_cmask = {}\n",
+	                          rt.attrib3.write_vrs_rate_hint_to_cmask ? "true" : "false"));
 	dst.push_back(fmt::format("\t dcc.max_uncompressed_block_size = 0x{:08x}\n",
 	                          rt.dcc.max_uncompressed_block_size));
 	dst.push_back(fmt::format("\t dcc.max_compressed_block_size   = 0x{:08x}\n",
 	                          rt.dcc.max_compressed_block_size));
-	dst.push_back(fmt::format("\t dcc.min_compressed_block_size   = 0x{:08x}\n",
-	                          rt.dcc.min_compressed_block_size));
 	dst.push_back(
 	    fmt::format("\t dcc.color_transform             = 0x{:08x}\n", rt.dcc.color_transform));
 	dst.push_back(fmt::format("\t dcc.overwrite_combiner_disable  = {}\n",
 	                          rt.dcc.overwrite_combiner_disable ? "true" : "false"));
-	dst.push_back(fmt::format("\t dcc.independent_64b_blocks      = {}\n",
-	                          rt.dcc.independent_64b_blocks ? "true" : "false"));
-	dst.push_back(fmt::format("\t dcc.independent_128b_blocks     = {}\n",
-	                          rt.dcc.independent_128b_blocks ? "true" : "false"));
+	dst.push_back(fmt::format("\t dcc.independent_block_size      = 0x{:02x}\n",
+	                          static_cast<uint8_t>(rt.dcc.independent_block_size)));
 	dst.push_back(fmt::format("\t data_write_on_dcc_clear_to_reg  = {}\n",
 	                          rt.dcc.data_write_on_dcc_clear_to_reg ? "true" : "false"));
 	dst.push_back(fmt::format("\t dcc.dcc_clear_key_enable        = {}\n",
 	                          rt.dcc.dcc_clear_key_enable ? "true" : "false"));
 	dst.push_back(fmt::format("\t cmask.addr                      = 0x{:016x}\n", rt.cmask.addr));
-	dst.push_back(fmt::format("\t cmask_slice.slice_minus1        = 0x{:08x}\n",
-	                          rt.cmask_slice.slice_minus1));
 	dst.push_back(fmt::format("\t fmask.addr                      = 0x{:016x}\n", rt.fmask.addr));
-	dst.push_back(fmt::format("\t fmask_slice.slice_minus1        = 0x{:08x}\n",
-	                          rt.fmask_slice.slice_minus1));
 	dst.push_back(
 	    fmt::format("\t clear_word0.word0               = 0x{:08x}\n", rt.clear_word0.word0));
 	dst.push_back(
 	    fmt::format("\t clear_word1.word1               = 0x{:08x}\n", rt.clear_word1.word1));
 	dst.push_back(
 	    fmt::format("\t dcc_addr.addr                   = 0x{:016x}\n", rt.dcc_addr.addr));
-	dst.push_back(fmt::format("\t size.width                      = 0x{:08x}\n", rt.size.width));
-	dst.push_back(fmt::format("\t size.height                     = 0x{:08x}\n", rt.size.height));
 
 	return dst;
 }
@@ -231,13 +205,8 @@ static bool RenderIsColorDimension(uint32_t dimension) {
 // NOLINTNEXTLINE(readability-function-cognitive-complexity)
 static void RtCheck(const HW::RenderTarget& rt) {
 	if (rt.base.addr != 0) {
-		// bool render_to_texture = (rt.attrib.tile_mode == 0x0d);
 		//  EXIT_NOT_IMPLEMENTED(rt.base_addr == 0);
 
-		EXIT_NOT_IMPLEMENTED(rt.pitch.pitch_div8_minus1 != 0);
-		EXIT_NOT_IMPLEMENTED(rt.pitch.fmask_pitch_div8_minus1 != 0);
-		EXIT_NOT_IMPLEMENTED(rt.slice.slice_div64_minus1 != 0);
-
 		EXIT_NOT_IMPLEMENTED(rt.view.base_array_slice_index > rt.view.last_array_slice_index);
 		if (rt.view.base_array_slice_index != 0x00000000 ||
 		    rt.view.last_array_slice_index != 0x00000000) {
@@ -257,11 +226,6 @@ static void RtCheck(const HW::RenderTarget& rt) {
 				logged = true;
 			}
 		}
-		const bool native_2x_one_frag =
-		    rt.info.fmask_compression_enable && !rt.info.fmask_data_compression_disable &&
-		    rt.info.cmask_fast_clear_enable && rt.fmask.addr != 0 && rt.cmask.addr != 0 &&
-		    rt.attrib.num_samples == 1 && rt.attrib.num_fragments == 1;
-		EXIT_NOT_IMPLEMENTED(rt.info.fmask_one_frag_mode && !native_2x_one_frag);
 		if (rt.info.fmask_compression_enable) {
 			EXIT_NOT_IMPLEMENTED(rt.attrib.num_samples == 0 && rt.attrib.num_fragments == 0);
 			static bool logged = false;
@@ -286,22 +250,6 @@ static void RtCheck(const HW::RenderTarget& rt) {
 				logged = true;
 			}
 		}
-		if (rt.info.cmask_is_linear != 0x00000000 || rt.info.cmask_addr_type != 0x00000000) {
-			static bool logged = false;
-			if (!logged) {
-				LOGF("RenderTarget: temporary: ignoring PS5 CMASK metadata fields "
-				     "is_linear=%u addr_type=%u\n",
-				     rt.info.cmask_is_linear, rt.info.cmask_addr_type);
-				logged = true;
-			}
-		}
-		if (rt.info.alt_tile_mode) {
-			static bool logged = false;
-			if (!logged) {
-				LOGF("RenderTarget: temporary: ignoring alternate tile mode flag\n");
-				logged = true;
-			}
-		}
 		if (rt.info.blend_bypass) {
 			static bool logged = false;
 			if (!logged) {
@@ -328,8 +276,6 @@ static void RtCheck(const HW::RenderTarget& rt) {
 				logged = true;
 			}
 		}
-		// EXIT_NOT_IMPLEMENTED(rt.tile_mode != 0x0000000a);
-		// EXIT_NOT_IMPLEMENTED(rt.fmask_tile_mode != 0x0000000a);
 		if (rt.attrib.num_samples != 0x00000000 || rt.attrib.num_fragments != 0x00000000) {
 			static bool logged = false;
 			if (!logged) {
@@ -364,34 +310,22 @@ static void RtCheck(const HW::RenderTarget& rt) {
 		if (!RenderIsColorDimension(rt.attrib3.dimension)) {
 			EXIT("unknown PS5 render-target dimension: 0x%08" PRIx32 "\n", rt.attrib3.dimension);
 		}
-		if (!rt.attrib3.cmask_pipe_aligned) {
+		if (!rt.attrib3.metadata_pipe_aligned) {
 			static bool logged = false;
 			if (!logged) {
-				LOGF("RenderTarget: temporary: accepting unaligned PS5 CMASK pipe flag\n");
-				logged = true;
-			}
-		}
-		if (!rt.attrib3.dcc_pipe_aligned) {
-			static bool logged = false;
-			if (!logged) {
-				LOGF("RenderTarget: temporary: accepting unaligned PS5 DCC pipe flag\n");
+				LOGF("RenderTarget: temporary: accepting unaligned PS5 metadata pipe flag\n");
 				logged = true;
 			}
 		}
+		EXIT_NOT_IMPLEMENTED(rt.attrib3.write_vrs_rate_hint_to_cmask);
 
 		// EXIT_NOT_IMPLEMENTED(rt.dcc_max_uncompressed_block_size != 0x00000002);
 		// EXIT_NOT_IMPLEMENTED(rt.dcc.max_compressed_block_size != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(rt.dcc.min_compressed_block_size != 0x00000000);
 		// EXIT_NOT_IMPLEMENTED(rt.dcc.color_transform != 0x00000000);
 		EXIT_NOT_IMPLEMENTED(rt.dcc.overwrite_combiner_disable != false);
-		// EXIT_NOT_IMPLEMENTED(rt.dcc.force_independent_blocks != false);
-		// EXIT_NOT_IMPLEMENTED(rt.dcc.independent_128b_blocks != false);
 		// EXIT_NOT_IMPLEMENTED(rt.dcc.data_write_on_dcc_clear_to_reg != false);
 		EXIT_NOT_IMPLEMENTED(rt.dcc.dcc_clear_key_enable != false);
-		if (rt.cmask.addr != 0x0000000000000000 || rt.cmask_slice.slice_minus1 != 0x00000000 ||
-		    rt.fmask.addr != 0x0000000000000000 ||
-		    (rt.fmask_slice.slice_minus1 != 0x00000000 &&
-		     rt.fmask_slice.slice_minus1 != rt.slice.slice_div64_minus1) ||
+		if (rt.cmask.addr != 0x0000000000000000 || rt.fmask.addr != 0x0000000000000000 ||
 		    rt.dcc_addr.addr != 0x0000000000000000) {
 			static bool logged = false;
 			if (!logged) {
@@ -401,9 +335,6 @@ static void RtCheck(const HW::RenderTarget& rt) {
 				logged = true;
 			}
 		}
-
-		EXIT_NOT_IMPLEMENTED(rt.size.width != 0);
-		EXIT_NOT_IMPLEMENTED(rt.size.height != 0);
 	}
 }
 
@@ -423,37 +354,20 @@ static void ZPrint(const char* func, const HW::DepthRenderTarget& z) {
 	     "\t stencil_info.htile_stencil_disabled   = %s\n"
 	     "\t stencil_info.expclear_enabled         = %s\n"
 	     "\t stencil_info.partially_resident       = %s\n"
-	     "\t depth_info.addr5_swizzle_mask         = 0x%08" PRIx32 "\n"
-	     "\t depth_info.array_mode                 = 0x%08" PRIx32 "\n"
-	     "\t depth_info.pipe_config                = 0x%08" PRIx32 "\n"
-	     "\t depth_info.bank_width                 = 0x%08" PRIx32 "\n"
-	     "\t depth_info.bank_height                = 0x%08" PRIx32 "\n"
-	     "\t depth_info.macro_tile_aspect          = 0x%08" PRIx32 "\n"
-	     "\t depth_info.num_banks                  = 0x%08" PRIx32 "\n"
 	     "\t depth_view.slice_start                = 0x%08" PRIx32 "\n"
 	     "\t depth_view.slice_max                  = 0x%08" PRIx32 "\n"
 	     "\t depth_view.current_mip_level          = 0x%02" PRIx8 "\n"
 	     "\t depth_view.depth_write_disable        = %s\n"
 	     "\t depth_view.stencil_write_disable      = %s\n"
-	     "\t htile_surface.linear                  = 0x%08" PRIx32 "\n"
-	     "\t htile_surface.full_cache              = 0x%08" PRIx32 "\n"
-	     "\t htile_surface.htile_uses_preload_win  = 0x%08" PRIx32 "\n"
-	     "\t htile_surface.preload                 = 0x%08" PRIx32 "\n"
-	     "\t htile_surface.prefetch_width          = 0x%08" PRIx32 "\n"
-	     "\t htile_surface.prefetch_height         = 0x%08" PRIx32 "\n"
-	     "\t htile_surface.dst_outside_zero_to_one = 0x%08" PRIx32 "\n"
 	     "\t z_read_base_addr                      = 0x%016" PRIx64 "\n"
 	     "\t stencil_read_base_addr                = 0x%016" PRIx64 "\n"
 	     "\t z_write_base_addr                     = 0x%016" PRIx64 "\n"
 	     "\t stencil_write_base_addr               = 0x%016" PRIx64 "\n"
-	     "\t pitch_div8_minus1                     = 0x%08" PRIx32 "\n"
-	     "\t height_div8_minus1                    = 0x%08" PRIx32 "\n"
-	     "\t slice_div64_minus1                    = 0x%08" PRIx32 "\n"
 	     "\t htile_data_base_addr                  = 0x%016" PRIx64 "\n"
-	     "\t width                                 = 0x%08" PRIx32 "\n"
-	     "\t height                                = 0x%08" PRIx32 "\n"
+	     "\t shading_rate_encoding                 = 0x%02" PRIx8 "\n"
 	     "\t size.x_max                            = 0x%04" PRIx16 "\n"
-	     "\t size.y_max                            = 0x%04" PRIx16 "\n",
+	     "\t size.y_max                            = 0x%04" PRIx16 "\n"
+	     "\t size.valid                            = %s\n",
 	     static_cast<uint32_t>(z.z_info.format), z.z_info.num_samples,
 	     static_cast<uint32_t>(z.z_info.texture_compatibility),
 	     z.z_info.htile_acceleration ? "true" : "false",
@@ -464,30 +378,27 @@ static void ZPrint(const char* func, const HW::DepthRenderTarget& z) {
 	     static_cast<uint32_t>(z.stencil_info.texture_compatibility),
 	     z.stencil_info.htile_stencil_disabled ? "true" : "false",
 	     z.stencil_info.expclear_enabled ? "true" : "false",
-	     z.stencil_info.partially_resident ? "true" : "false", z.depth_info.addr5_swizzle_mask,
-	     z.depth_info.array_mode, z.depth_info.pipe_config, z.depth_info.bank_width,
-	     z.depth_info.bank_height, z.depth_info.macro_tile_aspect, z.depth_info.num_banks,
-	     z.depth_view.slice_start, z.depth_view.slice_max, z.depth_view.current_mip_level,
+	     z.stencil_info.partially_resident ? "true" : "false", z.depth_view.slice_start,
+	     z.depth_view.slice_max, z.depth_view.current_mip_level,
 	     z.depth_view.depth_write_disable ? "true" : "false",
-	     z.depth_view.stencil_write_disable ? "true" : "false", z.htile_surface.linear,
-	     z.htile_surface.full_cache, z.htile_surface.htile_uses_preload_win,
-	     z.htile_surface.preload, z.htile_surface.prefetch_width, z.htile_surface.prefetch_height,
-	     z.htile_surface.dst_outside_zero_to_one, z.z_read_base_addr, z.stencil_read_base_addr,
-	     z.z_write_base_addr, z.stencil_write_base_addr, z.pitch_div8_minus1, z.height_div8_minus1,
-	     z.slice_div64_minus1, z.htile_data_base_addr, z.width, z.height, z.size.x_max,
-	     z.size.y_max);
+	     z.depth_view.stencil_write_disable ? "true" : "false", z.z_read_base_addr,
+	     z.stencil_read_base_addr, z.z_write_base_addr, z.stencil_write_base_addr,
+	     z.htile_data_base_addr, z.shading_rate_encoding, z.size.x_max, z.size.y_max,
+	     z.size.valid ? "true" : "false");
 }
 
 // NOLINTNEXTLINE(readability-function-cognitive-complexity)
 static void ZCheck(const HW::DepthRenderTarget& z, const HW::DepthControl& dc,
                    const HW::RenderControl& rc) {
-	const bool depth_active =
-	    dc.z_enable || dc.z_write_enable || dc.depth_bounds_enable || rc.depth_clear_enable;
-	const bool stencil_active = dc.stencil_enable || rc.stencil_clear_enable;
+	const bool depth_active = dc.z_enable || dc.z_write_enable || dc.depth_bounds_enable ||
+	                          rc.depth_clear_enable || rc.copy_depth_to_color;
+	const bool stencil_active =
+	    dc.stencil_enable || rc.stencil_clear_enable || rc.copy_stencil_to_color;
 	if (!depth_active && !stencil_active) {
 		return;
 	}
 
+	EXIT_NOT_IMPLEMENTED(rc.copy_depth_to_color || rc.copy_stencil_to_color);
 	EXIT_NOT_IMPLEMENTED(!z.z_info.HasValidTextureCompatibility());
 	EXIT_NOT_IMPLEMENTED(!z.stencil_info.HasValidTextureCompatibility());
 	if (z.z_info.format == Prospero::DepthFormat::kInvalid) {
@@ -524,21 +435,7 @@ static void ZCheck(const HW::DepthRenderTarget& z, const HW::DepthControl& dc,
 
 	if (z.z_info.format != Prospero::DepthFormat::kInvalid ||
 	    z.stencil_info.format != Prospero::StencilFormat::kInvalid) {
-
-		EXIT_NOT_IMPLEMENTED(z.depth_info.addr5_swizzle_mask != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.depth_info.array_mode != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.depth_info.pipe_config != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.depth_info.bank_width != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.depth_info.bank_height != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.depth_info.macro_tile_aspect != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.depth_info.num_banks != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.htile_surface.linear != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.htile_surface.full_cache != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.htile_surface.htile_uses_preload_win != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.htile_surface.preload != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.htile_surface.prefetch_width != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.htile_surface.prefetch_height != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.htile_surface.dst_outside_zero_to_one != 0x00000000);
+		EXIT_NOT_IMPLEMENTED(z.shading_rate_encoding != 0);
 
 		if (z.depth_view.current_mip_level != 0x00000000) {
 			static std::atomic<uint32_t> log_count {0};
@@ -560,23 +457,8 @@ static void ZCheck(const HW::DepthRenderTarget& z, const HW::DepthControl& dc,
 		EXIT_NOT_IMPLEMENTED(!z.depth_view.stencil_write_disable &&
 		                     z.stencil_read_base_addr != z.stencil_write_base_addr);
 		EXIT_NOT_IMPLEMENTED(!z.depth_view.depth_write_disable && z.z_write_base_addr == 0);
-		// EXIT_NOT_IMPLEMENTED(z.pitch_div8_minus1 != 0x000000ff);
-		// EXIT_NOT_IMPLEMENTED(z.height_div8_minus1 != 0x0000008f);
-		// EXIT_NOT_IMPLEMENTED(z.slice_div64_minus1 != 0x00008fff);
 		// EXIT_NOT_IMPLEMENTED(z.htile_data_base_addr == 0);
-		// EXIT_NOT_IMPLEMENTED(z.width != 0x00000780);
-		// EXIT_NOT_IMPLEMENTED(z.height != 0x00000438);
-
-		EXIT_NOT_IMPLEMENTED(z.width != 0);
-		EXIT_NOT_IMPLEMENTED(z.height != 0);
-		if (z.size.x_max == 0 || z.size.y_max == 0) {
-			static std::atomic<uint32_t> log_count {0};
-			if (log_count.fetch_add(1) < 16) {
-				LOGF("DepthTarget: temporary: inferring PS5 default depth extent x_max=0x%04" PRIx16
-				     ", y_max=0x%04" PRIx16 "\n",
-				     z.size.x_max, z.size.y_max);
-			}
-		}
+		EXIT_NOT_IMPLEMENTED(!z.size.valid);
 	}
 }
 
@@ -608,7 +490,7 @@ static void ClipCheck(const HW::ClipControl& c) {
 	// dx_linear_attr_clip_enable preserves linear (noperspective) attributes at clip-generated
 	// vertices, which Vulkan provides as part of clipping and interpolation.
 	EXIT_NOT_IMPLEMENTED(c.user_clip_planes != 0 || c.user_clip_plane_mode != 0 ||
-	                     c.vertex_kill_any || c.user_clip_plane_negate_y || c.clip_disable ||
+	                     c.vertex_kill_any || c.user_clip_plane_negate_y ||
 	                     c.user_clip_plane_cull_only || c.cull_on_clipping_error_disable ||
 	                     c.force_viewport_index_from_vs_enable);
 }
@@ -759,8 +641,6 @@ static void DPrint(const char* func, const HW::DepthControl& c, const HW::Stenci
 	     "\t backface_enable      = %s\n"
 	     "\t stencilfunc          = %" PRIu8 "\n"
 	     "\t stencilfunc_bf       = %" PRIu8 "\n"
-	     "\t color_writes_on_depth_fail_enable  = %s\n"
-	     "\t color_writes_on_depth_pass_disable = %s\n"
 	     "\t stencil_fail         = %" PRIu8 "\n"
 	     "\t stencil_zpass        = %" PRIu8 "\n"
 	     "\t stencil_zfail        = %" PRIu8 "\n"
@@ -777,12 +657,11 @@ static void DPrint(const char* func, const HW::DepthControl& c, const HW::Stenci
 	     "\t stencil_opval_bf     = %" PRIu8 "\n",
 	     c.stencil_enable ? "true" : "false", c.z_enable ? "true" : "false",
 	     c.z_write_enable ? "true" : "false", c.depth_bounds_enable ? "true" : "false", c.zfunc,
-	     c.backface_enable ? "true" : "false", c.stencilfunc, c.stencilfunc_bf,
-	     c.color_writes_on_depth_fail_enable ? "true" : "false",
-	     c.color_writes_on_depth_pass_disable ? "true" : "false", s.stencil_fail, s.stencil_zpass,
-	     s.stencil_zfail, s.stencil_fail_bf, s.stencil_zpass_bf, s.stencil_zfail_bf,
-	     sm.stencil_testval, sm.stencil_mask, sm.stencil_writemask, sm.stencil_opval,
-	     sm.stencil_testval_bf, sm.stencil_mask_bf, sm.stencil_writemask_bf, sm.stencil_opval_bf);
+	     c.backface_enable ? "true" : "false", c.stencilfunc, c.stencilfunc_bf, s.stencil_fail,
+	     s.stencil_zpass, s.stencil_zfail, s.stencil_fail_bf, s.stencil_zpass_bf,
+	     s.stencil_zfail_bf, sm.stencil_testval, sm.stencil_mask, sm.stencil_writemask,
+	     sm.stencil_opval, sm.stencil_testval_bf, sm.stencil_mask_bf, sm.stencil_writemask_bf,
+	     sm.stencil_opval_bf);
 }
 
 static void DCheck(const HW::DepthControl& c, const HW::StencilControl& s,
@@ -794,8 +673,6 @@ static void DCheck(const HW::DepthControl& c, const HW::StencilControl& s,
 	// Back-face stencil state is handled separately when enabled.
 	// EXIT_NOT_IMPLEMENTED(c.stencilfunc != 0);
 	// EXIT_NOT_IMPLEMENTED(c.stencilfunc_bf != 0);
-	EXIT_NOT_IMPLEMENTED(c.color_writes_on_depth_fail_enable != false);
-	EXIT_NOT_IMPLEMENTED(c.color_writes_on_depth_pass_disable != false);
 	// EXIT_NOT_IMPLEMENTED(s.stencil_fail != 0);
 	// EXIT_NOT_IMPLEMENTED(s.stencil_zpass != 0);
 	// EXIT_NOT_IMPLEMENTED(s.stencil_zfail != 0);
@@ -1202,7 +1079,7 @@ void hw_check(const CommandBuffer& buffer) {
 	AaCheck(aa, ac);
 	log_phase("done");
 
-	if (RenderTargetMaskHasBoundMrt(buffer)) {
+	if (graphics_debug_dump_enabled() && RenderTargetMaskHasBoundMrt(buffer)) {
 		LOGF("MRT render target mask: 0x%08" PRIx32 "\n", hw.GetRenderTargetMask());
 		for (uint32_t i = 0; i < 8; i++) {
 			const auto& mrt = hw.GetRenderTarget(i);

src/graphics/host_gpu/renderer/depthRenderTarget.cpp

@@ -108,9 +108,10 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, CommandBuffer&
 	const auto& sc          = hw.GetStencilControl();
 	const auto& sm          = hw.GetStencilMask();
 	const bool  has_stencil = z.stencil_info.format != Prospero::StencilFormat::kInvalid;
-	const bool  depth_active =
-	    dc.z_enable || dc.z_write_enable || dc.depth_bounds_enable || rc.depth_clear_enable;
-	const bool stencil_active = has_stencil && (dc.stencil_enable || rc.stencil_clear_enable);
+	const bool depth_active = dc.z_enable || dc.z_write_enable || dc.depth_bounds_enable ||
+	                          rc.depth_clear_enable || rc.copy_depth_to_color;
+	const bool stencil_active =
+	    has_stencil && (dc.stencil_enable || rc.stencil_clear_enable || rc.copy_stencil_to_color);
 	if (!depth_active && !stencil_active) {
 		return;
 	}
@@ -127,22 +128,14 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, CommandBuffer&
 	    !z.stencil_info.expclear_enabled && !z.stencil_info.partially_resident &&
 	    z.depth_view.slice_start == 0 && z.depth_view.slice_max == 0 &&
 	    z.depth_view.current_mip_level == 0 && !z.depth_view.depth_write_disable &&
-	    !z.depth_view.stencil_write_disable && z.depth_info.addr5_swizzle_mask == 0 &&
-	    z.depth_info.array_mode == 0 && z.depth_info.pipe_config == 0 &&
-	    z.depth_info.bank_width == 0 && z.depth_info.bank_height == 0 &&
-	    z.depth_info.macro_tile_aspect == 0 && z.depth_info.num_banks == 0 &&
-	    z.htile_surface.linear == 0 && z.htile_surface.full_cache == 0 &&
-	    z.htile_surface.htile_uses_preload_win == 0 && z.htile_surface.preload == 0 &&
-	    z.htile_surface.prefetch_width == 0 && z.htile_surface.prefetch_height == 0 &&
-	    z.htile_surface.dst_outside_zero_to_one == 0 && z.z_read_base_addr == 0 &&
-	    z.z_write_base_addr == 0 && z.stencil_read_base_addr == 0 &&
-	    z.stencil_write_base_addr == 0 && z.htile_data_base_addr == 0 &&
+	    !z.depth_view.stencil_write_disable && z.z_read_base_addr == 0 && z.z_write_base_addr == 0 &&
+	    z.stencil_read_base_addr == 0 && z.stencil_write_base_addr == 0 &&
+	    z.htile_data_base_addr == 0 &&
 	    // DB_DEPTH_SIZE_XY is independent state and may remain programmed after the attachment
 	    // formats and addresses are unbound. A zero encoding is the valid 1x1 value, so its
 	    // presence alone must not manufacture a depth attachment.
-	    !z.z_info.htile_acceleration && !z.width_height_valid && !z.pitch_height_valid &&
-	    z.size.x_max == 0 && z.size.y_max == 0 && z.pitch_div8_minus1 == 0 &&
-	    z.height_div8_minus1 == 0 && z.slice_div64_minus1 == 0 && z.width == 0 && z.height == 0;
+	    !z.z_info.htile_acceleration && z.shading_rate_encoding == 0 && z.size.x_max == 0 &&
+	    z.size.y_max == 0;
 	if (attachment_unbound) {
 		static std::atomic_bool logged = false;
 		if (!logged.exchange(true, std::memory_order_relaxed)) {
@@ -165,17 +158,12 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, CommandBuffer&
 			DepthFatal("invalid depth view: base=%u last=%u", z.depth_view.slice_start,
 			           z.depth_view.slice_max);
 	}
-	if (rc.copy_centroid || rc.copy_sample != 0 || z.z_info.expclear_enabled ||
+	if (rc.copy_depth_to_color || rc.copy_stencil_to_color || rc.copy_centroid ||
+	    rc.copy_sample != 0 || z.z_info.expclear_enabled ||
 	    z.stencil_info.expclear_enabled || z.z_info.partially_resident ||
 	    z.stencil_info.partially_resident || z.z_info.max_mip_level != 0 ||
-	    z.depth_view.current_mip_level != 0 || z.depth_info.addr5_swizzle_mask != 0 ||
-	    z.depth_info.array_mode != 0 || z.depth_info.pipe_config != 0 ||
-	    z.depth_info.bank_width != 0 || z.depth_info.bank_height != 0 ||
-	    z.depth_info.macro_tile_aspect != 0 || z.depth_info.num_banks != 0 ||
-	    z.htile_surface.linear != 0 || z.htile_surface.full_cache != 0 ||
-	    z.htile_surface.htile_uses_preload_win != 0 || z.htile_surface.preload != 0 ||
-	    z.htile_surface.prefetch_width != 0 || z.htile_surface.prefetch_height != 0 ||
-	    z.htile_surface.dst_outside_zero_to_one != 0 || z.z_read_base_addr == 0 ||
+	    z.depth_view.current_mip_level != 0 || z.shading_rate_encoding != 0 ||
+	    z.z_read_base_addr == 0 ||
 	    (!z.depth_view.depth_write_disable && z.z_write_base_addr != z.z_read_base_addr) ||
 	    (z.z_read_base_addr & 0xffffu) != 0 ||
 	    dc.zfunc > static_cast<uint8_t>(vk::CompareOp::eAlways)) {
@@ -189,10 +177,7 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, CommandBuffer&
 		    (z.stencil_read_base_addr & 0xffffu) != 0) {
 			DepthFatal("unsupported stencil attachment state");
 		}
-	} else if (z.stencil_read_base_addr != 0 || z.stencil_write_base_addr != 0 ||
-	           !htile_stencil_compat ||
-	           z.stencil_info.texture_compatibility !=
-	               Prospero::TextureCompatibleStencil::kDisable) {
+	} else if (z.stencil_read_base_addr != 0 || z.stencil_write_base_addr != 0) {
 		DepthFatal("stencil state without an active stencil attachment");
 	}
 	if (has_htile) {
@@ -203,18 +188,13 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, CommandBuffer&
 			DepthFatal("HTile clear tracking supports at most 32 slices");
 		}
 	}
-	const bool size_xy_valid = z.size.valid;
-	const bool wh_valid      = z.width_height_valid && z.width != 0 && z.height != 0;
-	if (!size_xy_valid && !wh_valid) {
+	if (!z.size.valid) {
 		DepthFatal("missing depth extent");
 	}
-	const uint32_t width  = size_xy_valid ? static_cast<uint32_t>(z.size.x_max) + 1u : z.width;
-	const uint32_t height = size_xy_valid ? static_cast<uint32_t>(z.size.y_max) + 1u : z.height;
-	if (width > 16384 || height > 16384 ||
-	    (size_xy_valid && wh_valid && (width != z.width || height != z.height)) ||
-	    (!z.pitch_height_valid &&
-	     (z.pitch_div8_minus1 != 0 || z.height_div8_minus1 != 0 || z.slice_div64_minus1 != 0))) {
-		DepthFatal("inconsistent depth extent or encoded layout");
+	const uint32_t width  = static_cast<uint32_t>(z.size.x_max) + 1u;
+	const uint32_t height = static_cast<uint32_t>(z.size.y_max) + 1u;
+	if (width > 16384 || height > 16384) {
+		DepthFatal("invalid depth extent");
 	}
 	const auto* policy = FindDepthFormatPolicy(z.z_info.format);
 	if (policy == nullptr) {
@@ -229,11 +209,6 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, CommandBuffer&
 	const auto     guest_format = policy->guest_format;
 	const uint32_t bytes        = policy->bytes_per_element;
 	const auto     pitch        = TileGetDepthPitch(width, bytes, z.z_info.num_samples);
-	if (z.pitch_height_valid && ((static_cast<uint64_t>(z.pitch_div8_minus1) + 1u) * 8u != pitch ||
-	                             (static_cast<uint64_t>(z.height_div8_minus1) + 1u) * 8u !=
-	                                 ((static_cast<uint64_t>(height) + 7u) & ~7ull))) {
-		DepthFatal("encoded depth pitch or height mismatch");
-	}
 	TileSizeAlign depth_size {};
 	TileSizeAlign stencil_size {};
 	TileSizeAlign htile_size {};
@@ -244,14 +219,6 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, CommandBuffer&
 	    (has_htile != (htile_size.align == 32768 && htile_size.size != 0))) {
 		DepthFatal("unsupported depth/stencil/HTile footprint");
 	}
-	if (z.pitch_height_valid &&
-	    (static_cast<uint64_t>(z.slice_div64_minus1) + 1u) * 64u != depth_size.size) {
-		DepthFatal("depth footprint mismatch: extent=%ux%u pitch=%u expected=0x%016" PRIx64
-		           " align=0x%016" PRIx64 " encoded_valid=%u encoded=0x%016" PRIx64,
-		           width, height, pitch, depth_size.size, depth_size.align,
-		           z.pitch_height_valid ? 1u : 0u,
-		           (static_cast<uint64_t>(z.slice_div64_minus1) + 1u) * 64u);
-	}
 	if (depth_size.size > UINT64_MAX / view.image_layers ||
 	    stencil_size.size > UINT64_MAX / view.image_layers ||
 	    htile_size.size > UINT64_MAX / view.image_layers) {
@@ -260,9 +227,9 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, CommandBuffer&
 	const auto depth_backing_size   = depth_size.size * view.image_layers;
 	const auto stencil_backing_size = stencil_size.size * view.image_layers;
 	const auto htile_backing_size   = htile_size.size * view.image_layers;
-	if (depth_backing_size > TRACKER_ADDRESS_SIZE - z.z_read_base_addr ||
-	    (has_stencil && stencil_backing_size > TRACKER_ADDRESS_SIZE - z.stencil_read_base_addr) ||
-	    (has_htile && htile_backing_size > TRACKER_ADDRESS_SIZE - z.htile_data_base_addr)) {
+	if (!GuestRange {z.z_read_base_addr, depth_backing_size}.Valid() ||
+	    (has_stencil && !GuestRange {z.stencil_read_base_addr, stencil_backing_size}.Valid()) ||
+	    (has_htile && !GuestRange {z.htile_data_base_addr, htile_backing_size}.Valid())) {
 		DepthFatal("layered depth backing range is invalid");
 	}
 	r.htile                   = has_htile;

src/graphics/host_gpu/renderer/depthRenderTarget.h

@@ -11,9 +11,11 @@
 
 namespace Libs::Graphics {
 
+// Without a stencil plane, Hi-Stencil fields are inactive. An active plane is compatible when
+// Hi-Stencil is disabled or HTile backing is present.
 inline constexpr bool depth_htile_stencil_acceleration_compatible(bool has_stencil, bool has_htile,
                                                                   bool htile_stencil_disabled) {
-	return htile_stencil_disabled || (has_stencil && has_htile);
+	return !has_stencil || htile_stencil_disabled || has_htile;
 }
 
 struct RenderDepthInfo {

src/graphics/host_gpu/renderer/image/image.cpp

@@ -73,10 +73,8 @@ namespace {
 	return usage;
 }
 
-void ValidateRange(GuestRange range, const char* name) {
-	if ((range.address == 0) != (range.size == 0) ||
-	    (range.address != 0 && (range.address >= TRACKER_ADDRESS_SIZE ||
-	                            range.size > TRACKER_ADDRESS_SIZE - range.address))) {
+void ValidateOptionalRange(GuestRange range, const char* name) {
+	if (!range.ValidOrEmpty()) {
 		EXIT("invalid %s image range: address=0x%016llx size=0x%016llx\n", name,
 		     static_cast<unsigned long long>(range.address),
 		     static_cast<unsigned long long>(range.size));
@@ -567,13 +565,13 @@ void Image::CopyMip(Image& source, uint32_t mip, uint32_t layer) {
 namespace ImageOps {
 
 void Validate(const ImageInfo& info) {
-	ValidateRange(info.data, "data");
-	ValidateRange(info.stencil, "stencil");
+	ValidateOptionalRange(info.data, "data");
+	ValidateOptionalRange(info.stencil, "stencil");
 
 	if (info.pixel_format == vk::Format::eUndefined) {
 		const bool metadata_empty =
-		    info.metadata.range.address == 0 && info.metadata.range.size == 0 &&
-		    info.metadata.kind == ImageMetadataKind::None && info.metadata.control == 0 &&
+		    info.metadata.range.Empty() && info.metadata.kind == ImageMetadataKind::None &&
+		    info.metadata.control == 0 &&
 		    info.metadata.compression == VideoOutCompression::Uncompressed &&
 		    !info.metadata.stencil_compressed;
 		if (info.data.Empty() || info.HasStencil() || !metadata_empty || info.extent.width == 0 ||
@@ -620,16 +618,14 @@ void Validate(const ImageInfo& info) {
 	}
 	switch (info.metadata.kind) {
 		case ImageMetadataKind::None:
-			if (info.metadata.range.address != 0 || info.metadata.range.size != 0 ||
-			    info.metadata.control != 0 ||
+			if (!info.metadata.range.Empty() || info.metadata.control != 0 ||
 			    info.metadata.compression != VideoOutCompression::Uncompressed ||
 			    info.metadata.stencil_compressed) {
 				EXIT("metadata-free image has metadata state\n");
 			}
 			break;
 		case ImageMetadataKind::Htile:
-			ValidateRange(info.metadata.range, "HTILE");
-			if (info.metadata.range.Empty() ||
+			if (!info.metadata.range.Valid() ||
 			    info.metadata.compression != VideoOutCompression::Uncompressed) {
 				EXIT("invalid HTILE metadata\n");
 			}

src/graphics/host_gpu/renderer/image/imageInfo.h

@@ -15,18 +15,6 @@
 
 namespace Libs::Graphics {
 
-struct GuestRange {
-	uint64_t address = 0;
-	uint64_t size    = 0;
-
-	[[nodiscard]] constexpr bool Empty() const noexcept { return address == 0 || size == 0; }
-	[[nodiscard]] constexpr bool Valid() const noexcept {
-		return !Empty() && address < TRACKER_ADDRESS_SIZE && size <= TRACKER_ADDRESS_SIZE - address;
-	}
-	[[nodiscard]] constexpr uint64_t End() const noexcept { return address + size; }
-	auto                             operator<=>(const GuestRange&) const = default;
-};
-
 enum class VideoOutCompression : uint8_t { Uncompressed, Dcc256_256_0, Dcc256_64_64, Unsupported };
 
 enum class ImageMetadataKind : uint8_t { None, Htile, Dcc };
@@ -443,6 +431,10 @@ IsSupportedDisplayRenderTargetTileMode(Prospero::TileMode tile_mode) noexcept {
 		return encoded <= 0.04045f ? encoded / 12.92f : std::pow((encoded + 0.055f) / 1.055f, 2.4f);
 	};
 	switch (format) {
+		// A single-plane float target carries its clear as raw float bits, the same encoding the
+		// depth decoder below uses. Without this the clear is discarded and the target keeps stale
+		// contents.
+		case vk::Format::eR32Sfloat: next.float32[0] = std::bit_cast<float>(packed); break;
 		case vk::Format::eR32Uint: next.uint32[0] = packed; break;
 		case vk::Format::eR32Sint: next.int32[0] = static_cast<int32_t>(packed); break;
 		case vk::Format::eR8G8B8A8Srgb:

src/graphics/host_gpu/renderer/pipeline/pipelineCache.cpp

@@ -116,7 +116,7 @@ struct PipelineCache::ProgramCache {
 	}
 
 	static constexpr std::size_t MaxStaticKeyWords =
-	    5 + ShaderVertexInputInfo::RES_MAX * 17;
+	    12 + ShaderVertexInputInfo::RES_MAX * 17;
 
 	template <typename InputInfo>
 	ShaderProgram Get(const ShaderParams& params, InputInfo& input_info) {
@@ -340,8 +340,23 @@ void PipelineCache::Save() {
 
 ShaderProgram PipelineCache::GetVertexProgram(const HW::VertexShaderInfo& regs,
                                               const HW::ShaderRegisters&  sh,
+                                              const HW::Context&          context,
                                               ShaderVertexInputInfo&      input_info) {
 	const auto params = PrepareProgram(regs, sh, input_info);
+	if (context.GetClipControl().clip_disable) {
+		const auto& viewport = context.GetScreenViewport().viewports[0];
+		const auto& limits   = m_graphics.GetPhysicalDeviceProperties().limits;
+		auto&       clip     = input_info.clip_space;
+		clip.scale[0]        = viewport.xscale;
+		clip.scale[1]        = viewport.yscale;
+		clip.offset[0]       = viewport.xoffset;
+		clip.offset[1]       = viewport.yoffset;
+		clip.half_extent[0] =
+		    static_cast<float>(std::min(limits.maxViewportDimensions[0], 16384u)) * 0.5f;
+		clip.half_extent[1] =
+		    static_cast<float>(std::min(limits.maxViewportDimensions[1], 16384u)) * 0.5f;
+		clip.enabled = true;
+	}
 	Common::LockGuard lock(m_mutex);
 	return m_program_cache->Get(params, input_info);
 }

src/graphics/host_gpu/renderer/pipeline/pipelineCache.h

@@ -130,6 +130,7 @@ public:
 
 	ShaderProgram GetVertexProgram(const HW::VertexShaderInfo& regs,
 	                               const HW::ShaderRegisters&  sh,
+	                               const HW::Context&          context,
 	                               ShaderVertexInputInfo&      input_info);
 	ShaderProgram
 	GetPixelProgram(const HW::PixelShaderInfo& regs, const HW::ShaderRegisters& sh,

src/graphics/host_gpu/renderer/renderCompute.cpp

@@ -54,7 +54,9 @@ bool RenderExecutor::TryConsumeComputeMetaClear(const ShaderComputeInputInfo& in
 	for (uint32_t i = 0; i < program.info.buffers.size(); i++) {
 		const auto& resource   = program.info.buffers[i];
 		const auto  descriptor = DecodeNativeDescriptor<ShaderBufferResource>(resources.buffers[i]);
-		if (!resource.written && cache.IsMeta(descriptor.Base48())) {
+		// A metadata resource that is also read is not proven to be a full overwrite. Execute it
+		// conservatively instead of replacing the dispatch with a coarse full-surface clear.
+		if (cache.IsMeta(descriptor.Base48()) && (!resource.written || resource.read)) {
 			return false;
 		}
 	}

src/graphics/host_gpu/renderer/renderDraw.cpp

@@ -331,10 +331,18 @@ static void SetGraphicsDynamicParams(const CommandBuffer& buffer, vk::CommandBuf
 	const auto final_scissor = calc_final_scissor(vp, ctx.GetScanModeControl(), framebuffer_extent);
 
 	vk::Viewport viewport {};
-	viewport.x        = vp.viewports[0].xoffset - vp.viewports[0].xscale;
-	viewport.y        = vp.viewports[0].yoffset - vp.viewports[0].yscale;
-	viewport.width    = vp.viewports[0].xscale * 2.0f;
-	viewport.height   = vp.viewports[0].yscale * 2.0f;
+	if (ctx.GetClipControl().clip_disable) {
+		const auto& limits = buffer.GetGraphics().GetPhysicalDeviceProperties().limits;
+		viewport.x         = 0.0f;
+		viewport.y         = 0.0f;
+		viewport.width     = static_cast<float>(std::min(limits.maxViewportDimensions[0], 16384u));
+		viewport.height    = static_cast<float>(std::min(limits.maxViewportDimensions[1], 16384u));
+	} else {
+		viewport.x      = vp.viewports[0].xoffset - vp.viewports[0].xscale;
+		viewport.y      = vp.viewports[0].yoffset - vp.viewports[0].yscale;
+		viewport.width  = vp.viewports[0].xscale * 2.0f;
+		viewport.height = vp.viewports[0].yscale * 2.0f;
+	}
 	viewport.minDepth = vp.viewports[0].zoffset;
 	viewport.maxDepth = vp.viewports[0].zscale + vp.viewports[0].zoffset;
 	vk_buffer.setViewport(0, 1, &viewport);
@@ -411,7 +419,7 @@ static void SetGraphicsDynamicParams(const CommandBuffer& buffer, vk::CommandBuf
 static bool DrawHasValidVertexShader(const HW::Shader& sh_ctx) {
 
 	const auto& vs = sh_ctx.GetVs();
-	return vs.gs_regs.chksum != 0 && ShaderAddressValid(vs.es_regs.data_addr);
+	return ShaderAddressValid(vs.es_regs.data_addr);
 }
 
 static bool PixelShaderHasDepthOrCoverageSideEffects(const HW::ShaderRegisters& sh_regs) {
@@ -443,7 +451,6 @@ static bool ShouldSkipGeShader(const CommandBuffer& buffer) {
 	};
 
 	const bool ps5_ngg_vertex_path = stages == 0x02002000 && vertex_info.es_regs.data_addr != 0 &&
-	                                 vertex_info.gs_regs.chksum != 0 &&
 	                                 sh_regs.m_vgtGsMaxVertOut == 0x00000000 &&
 	                                 is_known_gs_out_prim_type(sh_regs.m_vgtGsOutPrimType);
 
@@ -1014,8 +1021,8 @@ static void RefreshShaders(CommandBuffer& buffer, const DrawCallInfo& draw, bool
 		LogDrawPhase(draw.name, "GetVertexProgram");
 	}
 	auto& pipeline_cache = buffer.GetContext().GetPipelineCache();
-	state.vertex_program =
-	    pipeline_cache.GetVertexProgram(vertex_shader_info, shader_regs, state.vs_input_info);
+	state.vertex_program = pipeline_cache.GetVertexProgram(vertex_shader_info, shader_regs, ctx,
+	                                                         state.vs_input_info);
 
 	if (!state.ps_active) {
 		return;
@@ -1450,7 +1457,7 @@ void RenderExecutor::DrawAuto(uint64_t submit_id, CommandBuffer& buffer, uint32_
 		if (graphics_debug_dump_enabled()) {
 			LOGF("DrawIndexAuto: skipping rect-list draw with no VS param exports and PS inputs: "
 			     "ps_inputs=%u ps=0x%016" PRIx64 " es=0x%016" PRIx64 " gs=0x%016" PRIx64 "\n",
-			     state.ps_input_info.input_num, sh_ctx.GetPs().ps_regs.chksum,
+			     state.ps_input_info.input_num, sh_ctx.GetPs().ps_regs.data_addr,
 			     sh_ctx.GetVs().es_regs.data_addr, sh_ctx.GetVs().gs_regs.data_addr);
 		}
 		ResetBindings();

src/graphics/presentation/window/hostInput.cpp

@@ -29,6 +29,7 @@ struct ControlInfo {
 	uint32_t         button   = 0;
 	Controller::Axis axis     = Controller::Axis::AxisMax;
 	bool             positive = false;
+	float            touch_x  = 0.0f;
 };
 
 static constexpr std::array CONTROL_INFO = {
@@ -47,7 +48,10 @@ static constexpr std::array CONTROL_INFO = {
     ControlInfo {"Circle", Controller::PAD_BUTTON_CIRCLE},
     ControlInfo {"Cross", Controller::PAD_BUTTON_CROSS},
     ControlInfo {"Square", Controller::PAD_BUTTON_SQUARE},
-    ControlInfo {"TouchPad", Controller::PAD_BUTTON_TOUCH_PAD},
+    ControlInfo {"TouchPad", Controller::PAD_BUTTON_TOUCH_PAD, Controller::Axis::AxisMax, false,
+                 0.25f},
+    ControlInfo {"TouchPadRight", Controller::PAD_BUTTON_TOUCH_PAD, Controller::Axis::AxisMax,
+                 false, 0.75f},
     ControlInfo {"LeftStickLeft", 0, Controller::Axis::LeftX, false},
     ControlInfo {"LeftStickRight", 0, Controller::Axis::LeftX, true},
     ControlInfo {"LeftStickUp", 0, Controller::Axis::LeftY, false},
@@ -83,11 +87,6 @@ std::size_t ControlFromName(std::string_view name) {
 	return static_cast<std::size_t>(std::distance(CONTROL_INFO.begin(), info));
 }
 
-const ControlInfo& Info(std::size_t control) {
-	EXIT_IF(control >= CONTROL_INFO.size());
-	return CONTROL_INFO[control];
-}
-
 SDL_Keycode NormalizeKey(SDL_Keycode key) {
 	switch (key) {
 		case SDLK_RSHIFT: return SDLK_LSHIFT;
@@ -205,6 +204,10 @@ void SetButton(uint32_t button, bool down) {
 	}
 }
 
+void SetTouchPad(float x, bool down) {
+	Controller::ControllerTouchPad(Controller::HOST_INPUT_CONTROLLER_ID, 0, down, x, 0.5f);
+}
+
 uint32_t DefaultKeyboardButton(int key_code) {
 	switch (NormalizeKey(static_cast<SDL_Keycode>(key_code))) {
 		case SDLK_UP: return Controller::PAD_BUTTON_UP;
@@ -221,8 +224,6 @@ uint32_t DefaultKeyboardButton(int key_code) {
 		case SDLK_LCTRL: return Controller::PAD_BUTTON_R3;
 		case SDLK_RETURN:
 		case SDLK_RETURN2: return Controller::PAD_BUTTON_OPTIONS;
-		case SDLK_BACKSPACE:
-		case SDLK_TAB: return Controller::PAD_BUTTON_TOUCH_PAD;
 		default: return 0;
 	}
 }
@@ -235,7 +236,12 @@ struct StickKeys {
 };
 
 void SetStickAxis(Controller::Axis axis, bool negative, bool positive) {
-	const int value = negative == positive ? 128 : negative ? 0 : 255;
+	int value = 128;
+	if (negative && !positive) {
+		value = 0;
+	} else if (positive && !negative) {
+		value = 255;
+	}
 	Controller::ControllerAxis(Controller::HOST_INPUT_CONTROLLER_ID, axis, value);
 }
 
@@ -244,7 +250,11 @@ void SetControl(std::size_t control, bool down) {
 		return;
 	}
 
-	const auto& info = Info(control);
+	const auto& info = CONTROL_INFO[control];
+	if (info.button == Controller::PAD_BUTTON_TOUCH_PAD) {
+		SetTouchPad(info.touch_x, down);
+		return;
+	}
 	if (info.button != 0) {
 		SetButton(info.button, down);
 		return;
@@ -271,6 +281,8 @@ void DefaultKeyboardInput(int key_code, bool down) {
 	static StickKeys right;
 
 	switch (NormalizeKey(static_cast<SDL_Keycode>(key_code))) {
+		case SDLK_BACKSPACE: SetTouchPad(0.25f, down); return;
+		case SDLK_TAB: SetTouchPad(0.75f, down); return;
 		case SDLK_a:
 			left.left = down;
 			SetStickAxis(Controller::Axis::LeftX, left.left, left.right);

src/graphics/presentation/window/window.cpp

@@ -76,7 +76,6 @@ static uint32_t ControllerButtonToPadButton(int button) {
 		case SDL_CONTROLLER_BUTTON_B: return Controller::PAD_BUTTON_CIRCLE;
 		case SDL_CONTROLLER_BUTTON_X: return Controller::PAD_BUTTON_SQUARE;
 		case SDL_CONTROLLER_BUTTON_Y: return Controller::PAD_BUTTON_TRIANGLE;
-		case SDL_CONTROLLER_BUTTON_BACK: return Controller::PAD_BUTTON_TOUCH_PAD;
 		case SDL_CONTROLLER_BUTTON_START: return Controller::PAD_BUTTON_OPTIONS;
 		case SDL_CONTROLLER_BUTTON_LEFTSTICK: return Controller::PAD_BUTTON_L3;
 		case SDL_CONTROLLER_BUTTON_RIGHTSTICK: return Controller::PAD_BUTTON_R3;
@@ -86,6 +85,7 @@ static uint32_t ControllerButtonToPadButton(int button) {
 		case SDL_CONTROLLER_BUTTON_DPAD_DOWN: return Controller::PAD_BUTTON_DOWN;
 		case SDL_CONTROLLER_BUTTON_DPAD_LEFT: return Controller::PAD_BUTTON_LEFT;
 		case SDL_CONTROLLER_BUTTON_DPAD_RIGHT: return Controller::PAD_BUTTON_RIGHT;
+		case SDL_CONTROLLER_BUTTON_TOUCHPAD: return Controller::PAD_BUTTON_TOUCH_PAD;
 		default: return 0;
 	}
 }
@@ -242,7 +242,6 @@ static void GameEventKeyboard(WindowLoopState& game, const EventKeyboard& key) {
 #if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS || KYTY_PLATFORM == KYTY_PLATFORM_LINUX
 	if (key.down) {
 		switch (key.key_code) {
-			case SDLK_ESCAPE: game.need_exit = true; break;
 			case SDLK_SPACE: SetPause(game, !game.paused.load(std::memory_order_acquire)); break;
 			case SDLK_F1:
 				if (!key.repeat) {
@@ -705,6 +704,16 @@ void WindowContext::ProcessEvent(double time_s) {
 			break;
 		}
 
+		case SDL_CONTROLLERTOUCHPADDOWN:
+		case SDL_CONTROLLERTOUCHPADMOTION:
+		case SDL_CONTROLLERTOUCHPADUP:
+			if (event->ctouchpad.touchpad == 0) {
+				Controller::ControllerTouchPad(event->ctouchpad.which, event->ctouchpad.finger,
+				                               event->ctouchpad.type != SDL_CONTROLLERTOUCHPADUP,
+				                               event->ctouchpad.x, event->ctouchpad.y);
+			}
+			break;
+
 		case SDL_CONTROLLERDEVICEADDED:
 		case SDL_CONTROLLERDEVICEREMOVED:
 		case SDL_CONTROLLERDEVICEREMAPPED: {

src/graphics/shader/recompiler/ShaderRecompiler.cpp

@@ -30,14 +30,6 @@ namespace Libs::Graphics::ShaderRecompiler {
 
 namespace {
 
-bool ReadZeroMemory(void*, uint64_t, uint32_t* value) {
-	if (value == nullptr) {
-		return false;
-	}
-	*value = 0;
-	return true;
-}
-
 const char* GetDumpLabel(const CompileOptions& options) {
 	return options.dump_label != nullptr ? options.dump_label : "ShaderRecompiler";
 }
@@ -554,28 +546,20 @@ CompileResult Recompile(std::span<const uint32_t> code, const CompileOptions& op
 	switch (options.stage) {
 		case ShaderType::Vertex:
 			vertex = options.input_info.vertex;
-			if (vertex == nullptr) {
-				EXIT("vertex shader recompilation has no input metadata\n");
-			}
 			break;
 		case ShaderType::Pixel:
 			pixel = options.input_info.pixel;
-			if (pixel == nullptr) {
-				EXIT("pixel shader recompilation has no input metadata\n");
-			}
 			break;
 		case ShaderType::Compute:
 			compute = options.input_info.compute;
-			if (compute == nullptr) {
-				EXIT("compute shader recompilation has no input metadata\n");
-			}
 			break;
 		default: break;
 	}
 	EmbeddedFetchData embedded_fetch;
 	if (options.stage == ShaderType::Vertex && vertex->fetch_embedded) {
 		embedded_fetch = DetectEmbeddedVertexFetch(decoded, vertex, options.user_data_base,
-		                                           options.user_data_count, options.wave_size);
+		                                           static_cast<uint32_t>(options.user_data.size()),
+		                                           options.wave_size);
 		if (!embedded_fetch.loads.empty()) {
 			LOGF("%s embedded vertex fetch plan: detected=%" PRIu64 "\n", GetDumpLabel(options),
 			     static_cast<uint64_t>(embedded_fetch.loads.size()));
@@ -586,7 +570,7 @@ CompileResult Recompile(std::span<const uint32_t> code, const CompileOptions& op
 	    .wave_size           = options.wave_size,
 	    .shader_hash         = options.shader_hash,
 	    .user_data_base      = options.user_data_base,
-	    .user_data_count     = options.user_data_count,
+	    .user_data_count     = static_cast<uint32_t>(options.user_data.size()),
 	    .scratch_dwords      = options.scratch_dwords,
 	    .dispatcher_fallback = dispatcher_fallback,
 	    .cfg_failure_kind    = cfg.failure_kind,
@@ -634,32 +618,17 @@ CompileResult Recompile(std::span<const uint32_t> code, const CompileOptions& op
 		ir.info.vertex_offset_sgpr = embedded_fetch.vertex_offset_sgpr;
 	}
 
+	const IR::SrtRuntime runtime {
+	    .user_data                  = options.user_data,
+	    .shader_base                = reinterpret_cast<uint64_t>(code.data()),
+	    .read_memory                = options.read_memory,
+	    .userdata                   = options.read_memory_data,
+	    .read_specialization_memory = options.read_specialization_memory,
+	};
 	IR::ResourceSnapshot resources;
-	if (options.resource_snapshot != nullptr) {
-		resources = *options.resource_snapshot;
-	} else {
-		IR::SrtRuntime           runtime;
-		std::array<uint32_t, 64> zero_user_data {};
-		if (options.user_data == nullptr) {
-			for (uint32_t i = 2; i < zero_user_data.size(); i += 4) {
-				zero_user_data[i] = UINT32_MAX;
-			}
-		}
-		const auto* user_data =
-		    options.user_data != nullptr ? options.user_data : zero_user_data.data();
-		runtime.user_data   = std::span<const uint32_t>(user_data, options.user_data_count);
-		runtime.shader_base = options.shader_base != 0 ? options.shader_base
-		                                               : reinterpret_cast<uint64_t>(code.data());
-		runtime.read_memory = options.read_memory;
-		runtime.read_specialization_memory = options.read_specialization_memory;
-		if (runtime.read_memory == nullptr && options.user_data == nullptr) {
-			runtime.read_memory = ReadZeroMemory;
-		}
-		runtime.userdata         = options.read_memory_data;
-		if (!IR::MaterializeResources(ir, runtime, resources)) {
-			EXIT("shader resource materialization failed: stage=%s hash=0x%016" PRIx64 "\n",
-			     StageName(options.stage), options.shader_hash);
-		}
+	if (!IR::MaterializeResources(ir, runtime, resources)) {
+		EXIT("shader resource materialization failed: stage=%s hash=0x%016" PRIx64 "\n",
+		     StageName(options.stage), options.shader_hash);
 	}
 	IR::SpecializeResources(ir, resources);
 

src/graphics/shader/recompiler/ShaderRecompiler.h

@@ -15,19 +15,16 @@ struct CompileOptions {
 	ShaderType                  stage           = ShaderType::Compute;
 	uint32_t                    wave_size       = 64;
 	uint32_t                    user_data_base  = 0;
-	uint32_t                    user_data_count = 64;
 	uint32_t                    scratch_dwords  = 0;
 	uint64_t                    shader_hash     = 0;
-	uint64_t                    shader_base     = 0;
 	uint32_t                    push_constant_offset      = 0;
 	bool                        dump_ir                    = true;
 	bool                        early_dump                 = false;
 	const char*                 dump_label                 = nullptr;
-	const uint32_t*             user_data                  = nullptr;
+	std::span<const uint32_t>   user_data;
 	IR::SrtMemoryReader         read_memory                = nullptr;
 	IR::SrtMemoryReader         read_specialization_memory = nullptr;
 	void*                       read_memory_data           = nullptr;
-	const IR::ResourceSnapshot* resource_snapshot          = nullptr;
 	ShaderStageInputInfo        input_info;
 };
 

src/graphics/shader/recompiler/backend/spirv/spirvEmitterFlow.cpp

@@ -408,6 +408,41 @@ void EmitAuxPositionExport(ValueEmitContext& ctx, uint32_t data, const IR::Expor
 	}
 }
 
+uint32_t ConvertClipCoordinate(EmitterState& state, uint32_t coordinate, float scale,
+                               float offset, float half_extent) {
+	const auto window  = state.builder.AllocateId();
+	const auto biased  = state.builder.AllocateId();
+	const auto divided = state.builder.AllocateId();
+	const auto ndc     = state.builder.AllocateId();
+	state.builder.AddFunction(
+	    {OpFMul, TypeF32(state), window, coordinate, ConstantF32Value(state, scale)});
+	state.builder.AddFunction(
+	    {OpFAdd, TypeF32(state), biased, window, ConstantF32Value(state, offset)});
+	state.builder.AddFunction(
+	    {OpFDiv, TypeF32(state), divided, biased, ConstantF32Value(state, half_extent)});
+	state.builder.AddFunction(
+	    {OpFSub, TypeF32(state), ndc, divided, ConstantF32Value(state, 1.0f)});
+	return ndc;
+}
+
+uint32_t ConvertPositionToClipSpace(EmitterState& state, uint32_t position) {
+	const auto& transform = state.input_info.vertex->clip_space;
+	uint32_t    components[4] {};
+	for (uint32_t i = 0; i < 4; i++) {
+		components[i] = state.builder.AllocateId();
+		state.builder.AddFunction(
+		    {OpCompositeExtract, TypeF32(state), components[i], position, i});
+	}
+	components[0] = ConvertClipCoordinate(state, components[0], transform.scale[0],
+	                                      transform.offset[0], transform.half_extent[0]);
+	components[1] = ConvertClipCoordinate(state, components[1], transform.scale[1],
+	                                      transform.offset[1], transform.half_extent[1]);
+	const auto converted = state.builder.AllocateId();
+	state.builder.AddFunction({OpCompositeConstruct, TypeF32Vector(state, 4), converted,
+	                           components[0], components[1], components[2], components[3]});
+	return converted;
+}
+
 void EmitExport(ValueEmitContext& ctx, const IR::Inst& inst) {
 	auto&       state = ctx.state;
 	const auto& exp   = ctx.Export(inst);
@@ -467,6 +502,9 @@ void EmitExport(ValueEmitContext& ctx, const IR::Inst& inst) {
 			}
 		}
 		if (exp.kind == IR::ExportTargetKind::Position) {
+			if (state.stage == ShaderType::Vertex && state.input_info.vertex->clip_space.enabled) {
+				value = ConvertPositionToClipSpace(state, value);
+			}
 			const auto pointer = state.builder.AllocateId();
 			state.builder.AddFunction(
 			    {OpAccessChain, TypePointer(state, StorageClassOutput, TypeF32Vector(state, 4)),

src/graphics/shader/recompiler/backend/spirv/spirvEmitterHelpers.cpp

@@ -110,6 +110,12 @@ DppTargetLane EmitDppTargetLane(EmitterState& state, uint32_t control) {
 	if (control == 0x141u) {
 		return EmitDppMirrorTargetLane(state, subid, true);
 	}
+	if (control >= 0x160u && control <= 0x16fu) {
+		const auto target = state.builder.AllocateId();
+		state.builder.AddFunction(
+		    {OpBitwiseXor, TypeU32(state), target, subid, ConstantU32(state, control & 0xfu)});
+		return {target, EmitTrueBool(state)};
+	}
 	return {subid, EmitTrueBool(state)};
 }
 

src/graphics/shader/recompiler/frontend/decode/ScalarAluOps.cpp

@@ -45,6 +45,7 @@ constexpr OpcodeMap SOP1_OPCODE_LIST[] = {
     {0x10u, Opcode::S_BCNT1_I32_B64},
     {0x13u, Opcode::S_FF1_I32_B32},
     {0x14u, Opcode::S_FF1_I32_B64},
+    {0x15u, Opcode::S_FLBIT_I32_B32},
     {0x16u, Opcode::S_FLBIT_I32_B64},
     {0x1bu, Opcode::S_BITSET0_B32},
     {0x1du, Opcode::S_BITSET1_B32},

src/graphics/shader/recompiler/frontend/decode/ShaderDecoder.cpp

@@ -476,6 +476,7 @@ std::string InstructionToString(const Instruction& inst) {
 		case Opcode::S_ABS_I32:
 		case Opcode::S_BREV_B32:
 		case Opcode::S_BCNT1_I32_B32:
+		case Opcode::S_FLBIT_I32_B32:
 		case Opcode::S_FF1_I32_B32:
 		case Opcode::S_FF1_I32_B64:
 		case Opcode::S_NOT_B64:

src/graphics/shader/recompiler/frontend/decode/ShaderDecoder.h

@@ -45,6 +45,7 @@ enum class Opcode {
 	S_BCNT1_I32_B64,
 	S_FF1_I32_B32,
 	S_FF1_I32_B64,
+	S_FLBIT_I32_B32,
 	S_FLBIT_I32_B64,
 	S_BITREPLICATE_B64_B32,
 	S_QUADMASK_B64,
@@ -395,6 +396,7 @@ enum class Opcode {
 	V_CMP_GE_U32,
 	V_CMP_T_U32,
 	V_CMP_EQ_I64,
+	V_CMP_LT_U64,
 	V_CMP_EQ_U64,
 	V_CMP_GT_U64,
 	V_CMP_NE_U64,

src/graphics/shader/recompiler/frontend/decode/VectorAluOps.cpp

@@ -240,8 +240,9 @@ constexpr VopcOpcodeInfo VOPC_OPCODE_LIST[] = {
     {0xb5u, Opcode::V_CMPX_NE_I64, false}, {0xd1u, Opcode::V_CMPX_LT_U32},
     {0xd2u, Opcode::V_CMPX_EQ_U32},        {0xd3u, Opcode::V_CMPX_LE_U32},
     {0xd4u, Opcode::V_CMPX_GT_U32},        {0xd5u, Opcode::V_CMPX_NE_U32},
-    {0xd6u, Opcode::V_CMPX_GE_U32},        {0xe2u, Opcode::V_CMP_EQ_U64, false},
-    {0xe4u, Opcode::V_CMP_GT_U64, false},  {0xe5u, Opcode::V_CMP_NE_U64, false},
+    {0xd6u, Opcode::V_CMPX_GE_U32},        {0xe1u, Opcode::V_CMP_LT_U64, false},
+    {0xe2u, Opcode::V_CMP_EQ_U64, false},  {0xe4u, Opcode::V_CMP_GT_U64, false},
+    {0xe5u, Opcode::V_CMP_NE_U64, false},
     {0xf5u, Opcode::V_CMPX_NE_U64, false}, {0xc9u, Opcode::V_CMP_LT_F16},
     {0xcau, Opcode::V_CMP_EQ_F16},         {0xcbu, Opcode::V_CMP_LE_F16},
     {0xccu, Opcode::V_CMP_GT_F16},         {0xcdu, Opcode::V_CMP_LG_F16},
@@ -938,7 +939,7 @@ constexpr Vop2SdwaRule VOP2_SDWA_RULES[] = {
     {SdwaSelAll(), SdwaSelAll(), SdwaSelAll(), true, true},
     {SdwaSelFull(), SdwaSelAll(), SdwaSelAll(), false, false},
     {SdwaSelAll(), SdwaSelAll(), SdwaSelAll(), true, false},
-    {SdwaSelFull(), SdwaSelAll(), SdwaSelWords() | SdwaSelFull(), false, false},
+    {SdwaSelFull(), SdwaSelAll(), SdwaSelAll(), false, false},
     {SdwaSelFull(), SdwaSelAll(), SdwaSelAll(), false, false},
     {SdwaSelWords() | SdwaSelFull(), SdwaSelAll(), SdwaSelAll(), true, false},
 };

src/graphics/shader/recompiler/frontend/translate/Compare.cpp

@@ -12,7 +12,6 @@ void Translator::EmitCompareResult(const Decoder::Instruction& inst, IR::U1 valu
 	if (cmpx) {
 		const auto mask = BallotMask(masked);
 		ir.SetExec(masked);
-		ir.SetExecMaskTag(IR::U1(IR::Value(true)));
 		ir.SetExecLo(mask[0]);
 		ir.SetExecHi(mask[1]);
 		return;

src/graphics/shader/recompiler/frontend/translate/Control.cpp

@@ -24,28 +24,18 @@ Decoder::Operand ConditionOperand(Decoder::OperandKind kind) {
 
 void Translator::S_SAVEEXEC(const Decoder::Instruction& inst, IR::ValueOpcode operation,
                             bool negate_exec, bool negate_source, bool write_64) {
-	if (write_64) {
-		const auto exec_operand  = ConditionOperand(Decoder::OperandKind::ExecLo);
-		const auto old           = ReadScalarU64(exec_operand);
-		const auto bit_operation = operation == IR::ValueOpcode::LogicalAnd
-		                               ? IR::ValueOpcode::BitwiseAnd32
-		                               : IR::ValueOpcode::BitwiseOr32;
-		const auto result = BinaryScalarU64(old, ReadScalarU64(inst.src0), operation, bit_operation,
-		                                    negate_exec, negate_source, false);
-		WriteScalarU64(inst.dst, old);
-		WriteScalarU64(exec_operand, result);
-		ir.SetScc(ScalarU64NonZero(result));
-		return;
-	}
 	const auto old    = ir.GetExec();
 	const auto src    = ReadMask(inst.src0);
 	const auto lhs    = negate_exec ? ir.LogicalNot(old) : old;
 	const auto rhs    = negate_source ? ir.LogicalNot(src) : src;
 	const auto result = IR::U1(ir.Emit(operation, {lhs, rhs}));
-	WriteMask(inst.dst, old);
+	if (write_64) {
+		WriteMask64(inst.dst, old);
+	} else {
+		WriteMask(inst.dst, old);
+	}
 	const auto mask = BallotMask(result);
 	ir.SetExec(result);
-	ir.SetExecMaskTag(IR::U1(IR::Value(true)));
 	ir.SetExecLo(mask[0]);
 	ir.SetExecHi(mask[1]);
 	ir.SetScc(result);
@@ -201,21 +191,30 @@ void Translator::S_CSELECT_B32(const Decoder::Instruction& inst) {
 }
 
 void Translator::S_CSELECT_B64(const Decoder::Instruction& inst) {
-	const auto condition = ir.GetScc();
-	const auto lhs       = ReadScalarU64(inst.src0);
-	const auto rhs       = ReadScalarU64(inst.src1);
-	WriteScalarU64(inst.dst, SelectScalarU64(condition, lhs, rhs));
+	const auto condition     = ir.GetScc();
+	const auto lhs           = ReadU32Pair(inst.src0);
+	const auto rhs           = ReadU32Pair(inst.src1);
+	const auto selected_mask = IR::U1(
+	    ir.Emit(IR::ValueOpcode::SelectU1, {condition, ReadMask(inst.src0), ReadMask(inst.src1)}));
+	const auto selected_mask_valid =
+	    IR::U1(ir.Emit(IR::ValueOpcode::SelectU1,
+	                   {condition, ReadMaskValid(inst.src0), ReadMaskValid(inst.src1)}));
+	if (IsExecOrVcc(inst.dst)) {
+		WriteMask64(inst.dst, selected_mask);
+		return;
+	}
+	WriteU32Pair(inst.dst,
+	             {ir.Select(condition, lhs[0], rhs[0]), ir.Select(condition, lhs[1], rhs[1])});
+	if (inst.dst.kind == Decoder::OperandKind::Sgpr) {
+		const auto dst = static_cast<IR::ScalarReg>(inst.dst.reg);
+		ir.SetThreadBitScalarReg(dst, selected_mask);
+		ir.SetScalarMaskTag(dst, selected_mask_valid);
+	}
 }
 
 void Translator::MOV_B32(const Decoder::Instruction& inst, bool apply_float_modifiers) {
-	if (inst.src0.kind == Decoder::OperandKind::Sgpr &&
-	    inst.dst.kind == Decoder::OperandKind::Sgpr && inst.src0.reg == inst.dst.reg) {
-		return;
-	}
 	if (IsExecOrVcc(inst.src0) && IsExecOrVcc(inst.dst)) {
-		if (inst.src0.kind != inst.dst.kind) {
-			WriteRawU32(inst.dst, ReadRawU32(inst.src0));
-		}
+		WriteMask(inst.dst, ReadMask(inst.src0));
 	} else if (apply_float_modifiers && (inst.src0.negate || inst.src0.absolute)) {
 		WriteOperand(DestinationOperand(inst), ReadOperand(inst.src0, IR::Type::F32));
 	} else {
@@ -224,17 +223,50 @@ void Translator::MOV_B32(const Decoder::Instruction& inst, bool apply_float_modi
 }
 
 void Translator::S_MOV_B64(const Decoder::Instruction& inst) {
-	WriteScalarU64(inst.dst, ReadScalarU64(inst.src0));
+	if (IsExecOrVcc(inst.dst) || IsExecOrVcc(inst.src0)) {
+		WriteMask64(inst.dst, ReadMask(inst.src0));
+		return;
+	}
+	const bool scalar_copy =
+	    inst.dst.kind == Decoder::OperandKind::Sgpr && inst.src0.kind == Decoder::OperandKind::Sgpr;
+	IR::U1 source_mask;
+	IR::U1 source_mask_valid;
+	if (scalar_copy) {
+		source_mask       = ir.GetThreadBitScalarReg(static_cast<IR::ScalarReg>(inst.src0.reg));
+		source_mask_valid = ir.GetScalarMaskTag(static_cast<IR::ScalarReg>(inst.src0.reg));
+	}
+	WriteU32Pair(inst.dst, ReadU32Pair(inst.src0));
+	if (scalar_copy) {
+		ir.SetThreadBitScalarReg(static_cast<IR::ScalarReg>(inst.dst.reg), source_mask);
+		ir.SetScalarMaskTag(static_cast<IR::ScalarReg>(inst.dst.reg), source_mask_valid);
+	}
 }
 
 void Translator::S_WQM_B64(const Decoder::Instruction& inst) {
-	const auto source = ReadScalarU64(inst.src0);
-	const auto raw    = ExtractU64(
-	    IR::U64(ir.Emit(IR::ValueOpcode::WqmU64, {ir.ConstructU64(source.raw[0], source.raw[1])})));
-	const auto invocation = IR::U1(ir.Emit(IR::ValueOpcode::WqmMask, {source.invocation}));
-	const auto result     = MakeScalarU64Result(raw, invocation, source.mask_valid);
-	WriteScalarU64(inst.dst, result);
-	ir.SetScc(ScalarU64NonZero(result));
+	if (!IsExecOrVcc(inst.dst) && !IsExecOrVcc(inst.src0)) {
+		const auto mask_valid = ReadMaskValid(inst.src0);
+		const auto invocation_result =
+		    IR::U1(ir.Emit(IR::ValueOpcode::WqmMask, {ReadMask(inst.src0)}));
+		const auto result =
+		    IR::U64(ir.Emit(IR::ValueOpcode::WqmU64, {ReadOperand(inst.src0, IR::Type::U64)}));
+		WriteOperand(DestinationOperand(inst), result);
+		if (inst.dst.kind == Decoder::OperandKind::Sgpr) {
+			const auto dst = static_cast<IR::ScalarReg>(inst.dst.reg);
+			ir.SetThreadBitScalarReg(dst, invocation_result);
+			ir.SetScalarMaskTag(dst, mask_valid);
+			const auto raw_nonzero =
+			    IR::U1(ir.Emit(IR::ValueOpcode::INotEqual64, {result, IR::Value(uint64_t {0})}));
+			ir.SetScc(IR::U1(
+			    ir.Emit(IR::ValueOpcode::SelectU1, {mask_valid, invocation_result, raw_nonzero})));
+		} else {
+			ir.SetScc(
+			    IR::U1(ir.Emit(IR::ValueOpcode::INotEqual64, {result, IR::Value(uint64_t {0})})));
+		}
+		return;
+	}
+	const auto result = IR::U1(ir.Emit(IR::ValueOpcode::WqmMask, {ReadMask(inst.src0)}));
+	WriteMask64(inst.dst, result);
+	ir.SetScc(result);
 }
 
 void Translator::V_MOVRELS_B32(const Decoder::Instruction& inst) {

src/graphics/shader/recompiler/frontend/translate/Integer.cpp

@@ -77,15 +77,68 @@ bool Translator::PackedInteger16MinMax(const Decoder::Instruction& inst, IR::Val
 	return true;
 }
 
+IR::U1 Translator::U64MaskBinary(const Decoder::Instruction& inst, IR::ValueOpcode opcode,
+                                 bool negate_rhs, bool negate_result) {
+	const auto lhs = ReadMask(inst.src0);
+	auto       rhs = ReadMask(inst.src1);
+	if (negate_rhs) {
+		rhs = ir.LogicalNot(rhs);
+	}
+	auto result = IR::U1(ir.Emit(opcode, {lhs, rhs}));
+	return negate_result ? ir.LogicalNot(result) : result;
+}
+
 bool Translator::S_U64_MASK(const Decoder::Instruction& inst, IR::ValueOpcode logical_opcode,
                             IR::ValueOpcode bit_opcode, bool negate_rhs, bool negate_result,
                             bool unary) {
-	const auto lhs    = ReadScalarU64(inst.src0);
-	const auto result = unary ? NotScalarU64(lhs)
-	                          : BinaryScalarU64(lhs, ReadScalarU64(inst.src1), logical_opcode,
-	                                            bit_opcode, false, negate_rhs, negate_result);
-	WriteScalarU64(inst.dst, result);
-	ir.SetScc(ScalarU64NonZero(result));
+	const auto invocation_result =
+	    unary ? ir.LogicalNot(ReadMask(inst.src0))
+	          : U64MaskBinary(inst, logical_opcode, negate_rhs, negate_result);
+	const auto is_exec_or_vcc = [](const Decoder::Operand& operand) {
+		switch (operand.kind) {
+			case Decoder::OperandKind::ExecLo:
+			case Decoder::OperandKind::ExecHi:
+			case Decoder::OperandKind::VccLo:
+			case Decoder::OperandKind::VccHi: return true;
+			default: return false;
+		}
+	};
+	if (is_exec_or_vcc(inst.dst) || is_exec_or_vcc(inst.src0) ||
+	    (inst.src_count > 1u && is_exec_or_vcc(inst.src1))) {
+		WriteMask64(inst.dst, invocation_result);
+		ir.SetScc(invocation_result);
+		return true;
+	}
+
+	const auto lhs        = ReadU32Pair(inst.src0);
+	auto       mask_valid = ReadMaskValid(inst.src0);
+	if (inst.src_count > 1u) {
+		mask_valid = ir.LogicalAnd(mask_valid, ReadMaskValid(inst.src1));
+	}
+	std::array<IR::U32, 2> result;
+	if (unary) {
+		result = {ir.BitwiseNot(lhs[0]), ir.BitwiseNot(lhs[1])};
+	} else {
+		const auto rhs = ReadU32Pair(inst.src1);
+		for (uint32_t component = 0; component < 2u; component++) {
+			auto value = ir.Emit(
+			    bit_opcode, {lhs[component], negate_rhs ? IR::Value(ir.BitwiseNot(rhs[component]))
+			                                            : IR::Value(rhs[component])});
+			result[component] = negate_result ? ir.BitwiseNot(IR::U32(value)) : IR::U32(value);
+		}
+	}
+	WriteU32Pair(inst.dst, result);
+	if (inst.dst.kind == Decoder::OperandKind::Sgpr) {
+		const auto dst = static_cast<IR::ScalarReg>(inst.dst.reg);
+		ir.SetThreadBitScalarReg(dst, invocation_result);
+		ir.SetScalarMaskTag(dst, mask_valid);
+		const auto raw_nonzero =
+		    ir.INotEqual(ir.BitwiseOr(result[0], result[1]), IR::U32(IR::Value(0u)));
+		ir.SetScc(IR::U1(
+		    ir.Emit(IR::ValueOpcode::SelectU1, {mask_valid, invocation_result, raw_nonzero})));
+	} else {
+		ir.SetScc(ir.INotEqual(ir.BitwiseOr(result[0], result[1]), IR::U32(IR::Value(0u))));
+	}
 	return true;
 }
 

src/graphics/shader/recompiler/frontend/translate/Memory.cpp

@@ -55,14 +55,11 @@ uint32_t ResourceIndexFromOperand(const Decoder::Operand& operand) {
 	}
 }
 
-Decoder::Operand RawScalarLoadBase(const Decoder::Operand& operand) {
-	if (operand.kind != Decoder::OperandKind::M0) {
-		return operand;
+uint32_t RawScalarLoadBase(const Decoder::Operand& operand) {
+	if (operand.kind == Decoder::OperandKind::Sgpr) {
+		return operand.reg;
 	}
-	auto result = operand;
-	// The compressed SBASE value 62 decodes as code 124, but means NULL for S_LOAD.
-	result.kind = Decoder::OperandKind::Null;
-	return result;
+	return operand.kind == Decoder::OperandKind::VccLo ? 106u : 0u;
 }
 
 ResourceKind FlatSegmentResourceKind(uint32_t segment) {
@@ -322,9 +319,8 @@ Translator::AddressOperands Translator::ReadAddressOperands(const Decoder::Instr
 	return {GetAddressResource(low, high), low, high};
 }
 
-IR::Value Translator::GetScalarAddressResource(const Decoder::Operand& base) {
-	const auto raw = ReadU32Pair(base);
-	return GetAddressResource(raw[0], raw[1]);
+IR::Value Translator::GetScalarAddressResource(uint32_t base) {
+	return GetAddressResource(ReadScalarCode(base), ReadScalarCode(base + 1u));
 }
 
 IR::Value Translator::GetImageResource(const IR::MemoryInfo& memory) {

src/graphics/shader/recompiler/frontend/translate/Scalar.cpp

@@ -181,6 +181,7 @@ bool Translator::EmitScalar(const Decoder::Instruction& inst) {
 		case O::S_XNOR_B32:
 			return ComposedIntegerBinary(inst, IR::ValueOpcode::BitwiseXor32, false, true, true);
 		case O::S_FF1_I32_B64: return S_FF1_I32_B64(inst);
+		case O::S_FLBIT_I32_B32: return V_FFBH_32(inst, false);
 		case O::S_FLBIT_I32_B64: return S_FLBIT_I32_B64(inst);
 
 		case O::S_BITSET0_B32: return S_BITSET_B32(inst, false);

src/graphics/shader/recompiler/frontend/translate/Translate.cpp

@@ -151,8 +151,11 @@ IR::U32 Translator::ReadScalarCode(uint32_t code) {
 		case 106u: return ir.GetVccLo();
 		case 107u: return ir.GetVccHi();
 		case 124u: return ir.GetM0();
-		case 126u: return ir.GetExecLo();
-		case 127u: return ir.GetExecHi();
+		case 126u:
+		case 127u: {
+			const auto mask = BallotMask(ir.GetExec());
+			return mask[code - 126u];
+		}
 		default: return IR::U32(IR::Value(0u));
 	}
 }
@@ -312,23 +315,19 @@ void Translator::WriteRawU32(const Decoder::Operand& operand, IR::U32 value) {
 		case Decoder::OperandKind::VccLo:
 			ir.SetVccLo(IR::U32(value));
 			ir.SetVcc(ThreadBit(IR::U32(value)));
-			ir.SetVccMaskValidTag(IR::U1(IR::Value(false)));
 			break;
 		case Decoder::OperandKind::VccHi:
 			ir.SetVccHi(IR::U32(value));
 			ir.SetVcc(ThreadBit(ir.GetVccLo()));
-			ir.SetVccMaskValidTag(IR::U1(IR::Value(false)));
 			break;
 		case Decoder::OperandKind::M0: ir.SetM0(IR::U32(value)); break;
 		case Decoder::OperandKind::ExecLo:
 			ir.SetExecLo(IR::U32(value));
 			ir.SetExec(ThreadBit(IR::U32(value)));
-			ir.SetExecMaskTag(IR::U1(IR::Value(false)));
 			break;
 		case Decoder::OperandKind::ExecHi:
 			ir.SetExecHi(IR::U32(value));
 			ir.SetExec(ThreadBit(ir.GetExecLo()));
-			ir.SetExecMaskTag(IR::U1(IR::Value(false)));
 			break;
 		case Decoder::OperandKind::Scc:
 			ir.SetScc(ir.INotEqual(value, IR::U32(IR::Value(0u))));
@@ -373,7 +372,6 @@ void Translator::WriteOperand(const Decoder::Operand& operand, IR::Value value)
 			case Decoder::OperandKind::ExecHi: {
 				const auto mask = BallotMask(IR::U1(value));
 				ir.SetExec(IR::U1(value));
-				ir.SetExecMaskTag(IR::U1(IR::Value(true)));
 				ir.SetExecLo(mask[0]);
 				ir.SetExecHi(mask[1]);
 				return;
@@ -382,7 +380,6 @@ void Translator::WriteOperand(const Decoder::Operand& operand, IR::Value value)
 			case Decoder::OperandKind::VccHi: {
 				const auto mask = BallotMask(IR::U1(value));
 				ir.SetVcc(IR::U1(value));
-				ir.SetVccMaskValidTag(IR::U1(IR::Value(true)));
 				ir.SetVccLo(mask[0]);
 				ir.SetVccHi(mask[1]);
 				return;
@@ -583,13 +580,11 @@ void Translator::WriteU32Pair(const Decoder::Operand&       operand,
 	switch (operand.kind) {
 		case Decoder::OperandKind::ExecLo:
 			ir.SetExec(thread_bit);
-			ir.SetExecMaskTag(IR::U1(IR::Value(false)));
 			ir.SetExecLo(value[0]);
 			ir.SetExecHi(value[1]);
 			return;
 		case Decoder::OperandKind::VccLo:
 			ir.SetVcc(thread_bit);
-			ir.SetVccMaskValidTag(IR::U1(IR::Value(false)));
 			ir.SetVccLo(value[0]);
 			ir.SetVccHi(value[1]);
 			return;
@@ -600,87 +595,6 @@ void Translator::WriteU32Pair(const Decoder::Operand&       operand,
 	WriteRawU32(OffsetOperand(operand, 1), value[1]);
 }
 
-Translator::ScalarU64 Translator::ReadScalarU64(const Decoder::Operand& operand) {
-	return {ReadU32Pair(operand), ReadMask(operand), ReadMaskValid(operand)};
-}
-
-Translator::ScalarU64 Translator::MakeScalarU64Mask(IR::U1 value) {
-	return {BallotMask(value), value, IR::U1(IR::Value(true))};
-}
-
-Translator::ScalarU64 Translator::MakeScalarU64Result(const std::array<IR::U32, 2>& raw,
-                                                      IR::U1 invocation, IR::U1 mask_valid) {
-	return {raw,
-	        IR::U1(ir.Emit(IR::ValueOpcode::SelectU1, {mask_valid, invocation, ThreadBit(raw[0])})),
-	        mask_valid};
-}
-
-Translator::ScalarU64 Translator::SelectScalarU64(IR::U1 condition, const ScalarU64& lhs,
-                                                  const ScalarU64& rhs) {
-	return {
-	    {ir.Select(condition, lhs.raw[0], rhs.raw[0]),
-	     ir.Select(condition, lhs.raw[1], rhs.raw[1])},
-	    IR::U1(ir.Emit(IR::ValueOpcode::SelectU1, {condition, lhs.invocation, rhs.invocation})),
-	    IR::U1(ir.Emit(IR::ValueOpcode::SelectU1, {condition, lhs.mask_valid, rhs.mask_valid}))};
-}
-
-Translator::ScalarU64 Translator::NotScalarU64(const ScalarU64& value) {
-	return MakeScalarU64Result({ir.BitwiseNot(value.raw[0]), ir.BitwiseNot(value.raw[1])},
-	                           ir.LogicalNot(value.invocation), value.mask_valid);
-}
-
-Translator::ScalarU64 Translator::BinaryScalarU64(const ScalarU64& lhs, const ScalarU64& rhs,
-                                                  IR::ValueOpcode logical_opcode,
-                                                  IR::ValueOpcode bit_opcode, bool negate_lhs,
-                                                  bool negate_rhs, bool negate_result) {
-	auto lhs_mask = negate_lhs ? ir.LogicalNot(lhs.invocation) : lhs.invocation;
-	auto rhs_mask = negate_rhs ? ir.LogicalNot(rhs.invocation) : rhs.invocation;
-	auto mask     = IR::U1(ir.Emit(logical_opcode, {lhs_mask, rhs_mask}));
-	if (negate_result) {
-		mask = ir.LogicalNot(mask);
-	}
-	std::array<IR::U32, 2> raw;
-	for (uint32_t component = 0; component < raw.size(); component++) {
-		auto lhs_raw   = negate_lhs ? ir.BitwiseNot(lhs.raw[component]) : lhs.raw[component];
-		auto rhs_raw   = negate_rhs ? ir.BitwiseNot(rhs.raw[component]) : rhs.raw[component];
-		raw[component] = IR::U32(ir.Emit(bit_opcode, {lhs_raw, rhs_raw}));
-		if (negate_result) {
-			raw[component] = ir.BitwiseNot(raw[component]);
-		}
-	}
-	return MakeScalarU64Result(raw, mask, ir.LogicalAnd(lhs.mask_valid, rhs.mask_valid));
-}
-
-void Translator::WriteScalarU64(const Decoder::Operand& operand, const ScalarU64& value) {
-	WriteU32Pair(operand, value.raw);
-	switch (operand.kind) {
-		case Decoder::OperandKind::ExecLo:
-		case Decoder::OperandKind::ExecHi:
-			ir.SetExec(value.invocation);
-			ir.SetExecMaskTag(value.mask_valid);
-			break;
-		case Decoder::OperandKind::VccLo:
-		case Decoder::OperandKind::VccHi:
-			ir.SetVcc(value.invocation);
-			ir.SetVccMaskValidTag(value.mask_valid);
-			break;
-		case Decoder::OperandKind::Sgpr: {
-			const auto reg = static_cast<IR::ScalarReg>(operand.reg);
-			ir.SetThreadBitScalarReg(reg, value.invocation);
-			ir.SetScalarMaskTag(reg, value.mask_valid);
-			break;
-		}
-		default: break;
-	}
-}
-
-IR::U1 Translator::ScalarU64NonZero(const ScalarU64& value) {
-	const auto raw_nonzero =
-	    ir.INotEqual(ir.BitwiseOr(value.raw[0], value.raw[1]), IR::U32(IR::Value(0u)));
-	return IR::U1(
-	    ir.Emit(IR::ValueOpcode::SelectU1, {value.mask_valid, value.invocation, raw_nonzero}));
-}
-
 IR::U1 Translator::ThreadBit(IR::U32 low) {
 	return ir.INotEqual(low, IR::U32(IR::Value(0u)));
 }
@@ -736,9 +650,9 @@ IR::U1 Translator::ReadMaskValid(const Decoder::Operand& operand) {
 		case Decoder::OperandKind::Sgpr:
 			return ir.GetScalarMaskTag(static_cast<IR::ScalarReg>(operand.reg));
 		case Decoder::OperandKind::ExecLo:
-		case Decoder::OperandKind::ExecHi: return ir.GetExecMaskTag();
+		case Decoder::OperandKind::ExecHi:
 		case Decoder::OperandKind::VccLo:
-		case Decoder::OperandKind::VccHi: return ir.GetVccMaskValidTag();
+		case Decoder::OperandKind::VccHi:
 		case Decoder::OperandKind::VccZ:
 		case Decoder::OperandKind::ExecZ:
 		case Decoder::OperandKind::Scc: return IR::U1(IR::Value(true));
@@ -771,7 +685,6 @@ void Translator::WriteMask(const Decoder::Operand& operand, IR::U1 value) {
 		case Decoder::OperandKind::ExecHi: {
 			const auto mask = BallotMask(value);
 			ir.SetExec(value);
-			ir.SetExecMaskTag(IR::U1(IR::Value(true)));
 			ir.SetExecLo(mask[0]);
 			ir.SetExecHi(mask[1]);
 			return;
@@ -780,7 +693,6 @@ void Translator::WriteMask(const Decoder::Operand& operand, IR::U1 value) {
 		case Decoder::OperandKind::VccHi: {
 			const auto mask = BallotMask(value);
 			ir.SetVcc(value);
-			ir.SetVccMaskValidTag(IR::U1(IR::Value(true)));
 			ir.SetVccLo(mask[0]);
 			ir.SetVccHi(mask[1]);
 			return;
@@ -794,7 +706,25 @@ void Translator::WriteMask(const Decoder::Operand& operand, IR::U1 value) {
 }
 
 void Translator::WriteMask64(const Decoder::Operand& operand, IR::U1 value) {
-	WriteScalarU64(operand, MakeScalarU64Mask(value));
+	if (operand.kind != Decoder::OperandKind::Sgpr) {
+		WriteMask(operand, value);
+		return;
+	}
+	const auto reg  = static_cast<IR::ScalarReg>(operand.reg);
+	const auto mask = BallotMask(value);
+	ir.SetThreadBitScalarReg(reg, value);
+	ir.SetScalarMaskTag(reg, IR::U1(IR::Value(true)));
+	if (IR::RegIndex(reg) > 0u) {
+		ir.SetScalarMaskTag(static_cast<IR::ScalarReg>(IR::RegIndex(reg) - 1u),
+		                    IR::U1(IR::Value(false)));
+	}
+	ir.SetScalarReg(reg, mask[0]);
+	if (IR::RegIndex(reg) + 1u < IR::NumScalarRegs) {
+		const auto high = static_cast<IR::ScalarReg>(IR::RegIndex(reg) + 1u);
+		ir.SetScalarReg(high, mask[1]);
+		ir.SetThreadBitScalarReg(high, IR::U1(IR::Value(false)));
+		ir.SetScalarMaskTag(high, IR::U1(IR::Value(false)));
+	}
 }
 
 void Translator::WriteCompareResult(const Decoder::Operand& operand, IR::U1 value) {
@@ -1095,7 +1025,6 @@ IR::Program TranslateProgram(const Decoder::Program& decoded, const CFG::Graph&
 			entry_ir.SetScalarMaskTag(reg, IR::U1(IR::Value(false)));
 		}
 		entry_ir.SetExec(IR::U1(IR::Value(true)));
-		entry_ir.SetExecMaskTag(IR::U1(IR::Value(true)));
 		entry_ir.SetExecLo(IR::U32(IR::Value(1u)));
 		entry_ir.SetExecHi(IR::U32(IR::Value(0u)));
 		if (options.stage == ShaderType::Compute) {

src/graphics/shader/recompiler/frontend/translate/Translator.h

@@ -19,14 +19,6 @@ public:
 	void AddBranchCondition(const CFG::BasicBlock& source, IR::BlockInfo& info);
 
 private:
-	struct ScalarU64 {
-		std::array<IR::U32, 2> raw;
-		// Raw words and Kyty's per-invocation mask model remain independent.
-		// mask_valid selects invocation for logical consumers and SCC.
-		IR::U1 invocation;
-		IR::U1 mask_valid;
-	};
-
 	Decoder::Operand SourceAt(const Decoder::Instruction& inst, uint32_t index);
 	Decoder::Operand DestinationOperand(const Decoder::Instruction& inst);
 	Decoder::Operand OffsetOperand(const Decoder::Operand& operand, uint32_t offset);
@@ -56,25 +48,14 @@ private:
 	IR::U32 ReadU16AsU32(const Decoder::Operand& operand, bool sign_extend);
 	IR::U32 ReadF16LaneBits(const Decoder::Operand& operand, bool high_lane);
 	std::array<IR::U32, 2> ExtractU64(IR::U64 value);
-	void      WriteU32Pair(const Decoder::Operand& operand, const std::array<IR::U32, 2>& value);
-	ScalarU64 ReadScalarU64(const Decoder::Operand& operand);
-	ScalarU64 MakeScalarU64Mask(IR::U1 value);
-	ScalarU64 MakeScalarU64Result(const std::array<IR::U32, 2>& raw, IR::U1 invocation,
-	                              IR::U1 mask_valid);
-	ScalarU64 SelectScalarU64(IR::U1 condition, const ScalarU64& lhs, const ScalarU64& rhs);
-	ScalarU64 NotScalarU64(const ScalarU64& value);
-	ScalarU64 BinaryScalarU64(const ScalarU64& lhs, const ScalarU64& rhs,
-	                          IR::ValueOpcode logical_opcode, IR::ValueOpcode bit_opcode,
-	                          bool negate_lhs, bool negate_rhs, bool negate_result);
-	void      WriteScalarU64(const Decoder::Operand& operand, const ScalarU64& value);
-	IR::U1    ScalarU64NonZero(const ScalarU64& value);
-	IR::U1    ReadCondition(const Decoder::Operand& operand);
-	IR::U32   ConditionBit(const Decoder::Operand& operand);
-	IR::U1    ReadMask(const Decoder::Operand& operand);
-	IR::U1    ReadMaskValid(const Decoder::Operand& operand);
-	void      WriteMask(const Decoder::Operand& operand, IR::U1 value);
-	void      WriteMask64(const Decoder::Operand& operand, IR::U1 value);
-	void      WriteCompareResult(const Decoder::Operand& operand, IR::U1 value);
+	void    WriteU32Pair(const Decoder::Operand& operand, const std::array<IR::U32, 2>& value);
+	IR::U1  ReadCondition(const Decoder::Operand& operand);
+	IR::U32 ConditionBit(const Decoder::Operand& operand);
+	IR::U1  ReadMask(const Decoder::Operand& operand);
+	IR::U1  ReadMaskValid(const Decoder::Operand& operand);
+	void    WriteMask(const Decoder::Operand& operand, IR::U1 value);
+	void    WriteMask64(const Decoder::Operand& operand, IR::U1 value);
+	void    WriteCompareResult(const Decoder::Operand& operand, IR::U1 value);
 
 	IR::MemoryFlags AddMemoryInfo(const IR::MemoryInfo& memory, uint32_t pc);
 	IR::ExportFlags AddExportInfo(const Decoder::Instruction& inst);
@@ -92,7 +73,7 @@ private:
 	IR::U32         GetResourceDword(uint32_t index, uint32_t dword);
 	IR::Value       GetBufferResource(const IR::MemoryInfo& memory);
 	IR::Value       GetAddressResource(IR::Value low, IR::Value high);
-	IR::Value       GetScalarAddressResource(const Decoder::Operand& base);
+	IR::Value       GetScalarAddressResource(uint32_t base);
 	IR::Value       GetImageResource(const IR::MemoryInfo& memory);
 	IR::Value       GetSamplerResource(const IR::MemoryInfo& memory);
 	IR::Value     MakeImageAddress(const Decoder::Instruction& inst, const Decoder::Operand& base);
@@ -194,6 +175,8 @@ private:
 	bool PackedInteger16MinMax(const Decoder::Instruction& inst, IR::ValueOpcode opcode, bool sign);
 	bool S_U64_MASK(const Decoder::Instruction& inst, IR::ValueOpcode logical_opcode,
 	                IR::ValueOpcode bit_opcode, bool negate_rhs, bool negate_result, bool unary);
+	IR::U1 U64MaskBinary(const Decoder::Instruction& inst, IR::ValueOpcode opcode, bool negate_rhs,
+	                     bool negate_result);
 	bool SimpleInteger(const Decoder::Instruction& inst, IR::ValueOpcode opcode, IR::Type type,
 	                   bool reverse, bool mask_shift_count, bool update_scc);
 	bool ComposedIntegerBinary(const Decoder::Instruction& inst, IR::ValueOpcode opcode,

src/graphics/shader/recompiler/frontend/translate/Vector.cpp

@@ -99,6 +99,9 @@ bool Translator::EmitVector(const Decoder::Instruction& inst) {
 		case O::V_CMP_EQ_U64:
 			EmitIntegerCompare(inst, IR::ValueOpcode::IEqual64, IR::Type::U64, false, false);
 			return true;
+		case O::V_CMP_LT_U64:
+			EmitIntegerCompare(inst, IR::ValueOpcode::ULessThan64, IR::Type::U64, false, false);
+			return true;
 		case O::V_CMP_GT_U64:
 			EmitIntegerCompare(inst, IR::ValueOpcode::UGreaterThan64, IR::Type::U64, false, false);
 			return true;

src/graphics/shader/recompiler/ir/IREmitter.cpp

@@ -76,14 +76,6 @@ void IREmitter::SetExec(U1 value) {
 	Emit(ValueOpcode::SetExec, {value});
 }
 
-U1 IREmitter::GetExecMaskTag() {
-	return U1(Emit(ValueOpcode::GetExecMaskTag));
-}
-
-void IREmitter::SetExecMaskTag(U1 value) {
-	Emit(ValueOpcode::SetExecMaskTag, {value});
-}
-
 U32 IREmitter::GetExecLo() {
 	return U32(Emit(ValueOpcode::GetExecLo));
 }
@@ -108,14 +100,6 @@ void IREmitter::SetVcc(U1 value) {
 	Emit(ValueOpcode::SetVcc, {value});
 }
 
-U1 IREmitter::GetVccMaskValidTag() {
-	return U1(Emit(ValueOpcode::GetVccMaskValidTag));
-}
-
-void IREmitter::SetVccMaskValidTag(U1 value) {
-	Emit(ValueOpcode::SetVccMaskValidTag, {value});
-}
-
 U32 IREmitter::GetVccLo() {
 	return U32(Emit(ValueOpcode::GetVccLo));
 }

src/graphics/shader/recompiler/ir/IREmitter.h

@@ -39,16 +39,12 @@ public:
 	void SetScc(U1 value);
 	U1   GetExec();
 	void SetExec(U1 value);
-	U1   GetExecMaskTag();
-	void SetExecMaskTag(U1 value);
 	U32  GetExecLo();
 	void SetExecLo(U32 value);
 	U32  GetExecHi();
 	void SetExecHi(U32 value);
 	U1   GetVcc();
 	void SetVcc(U1 value);
-	U1   GetVccMaskValidTag();
-	void SetVccMaskValidTag(U1 value);
 	U32  GetVccLo();
 	void SetVccLo(U32 value);
 	U32  GetVccHi();

src/graphics/shader/recompiler/ir/Program.cpp

@@ -33,16 +33,12 @@ bool IsRegisterStatePseudo(ValueOpcode opcode) {
 		case ValueOpcode::SetScc:
 		case ValueOpcode::GetExec:
 		case ValueOpcode::SetExec:
-		case ValueOpcode::GetExecMaskTag:
-		case ValueOpcode::SetExecMaskTag:
 		case ValueOpcode::GetExecLo:
 		case ValueOpcode::SetExecLo:
 		case ValueOpcode::GetExecHi:
 		case ValueOpcode::SetExecHi:
 		case ValueOpcode::GetVcc:
 		case ValueOpcode::SetVcc:
-		case ValueOpcode::GetVccMaskValidTag:
-		case ValueOpcode::SetVccMaskValidTag:
 		case ValueOpcode::GetVccLo:
 		case ValueOpcode::SetVccLo:
 		case ValueOpcode::GetVccHi:

src/graphics/shader/recompiler/ir/opcodes/ValueOpcodes.inc

@@ -23,16 +23,12 @@ VALUE_OPCODE(GetScc, U1)
 VALUE_OPCODE(SetScc, Void, U1)
 VALUE_OPCODE(GetExec, U1)
 VALUE_OPCODE(SetExec, Void, U1)
-VALUE_OPCODE(GetExecMaskTag, U1)
-VALUE_OPCODE(SetExecMaskTag, Void, U1)
 VALUE_OPCODE(GetExecLo, U32)
 VALUE_OPCODE(SetExecLo, Void, U32)
 VALUE_OPCODE(GetExecHi, U32)
 VALUE_OPCODE(SetExecHi, Void, U32)
 VALUE_OPCODE(GetVcc, U1)
 VALUE_OPCODE(SetVcc, Void, U1)
-VALUE_OPCODE(GetVccMaskValidTag, U1)
-VALUE_OPCODE(SetVccMaskValidTag, Void, U1)
 VALUE_OPCODE(GetVccLo, U32)
 VALUE_OPCODE(SetVccLo, Void, U32)
 VALUE_OPCODE(GetVccHi, U32)

src/graphics/shader/recompiler/ir/passes/ResourceMaterialization.cpp

@@ -61,7 +61,8 @@ bool NullImageDescriptor(const DescriptorValue& descriptor) {
 bool ValidImageDescriptor(const DescriptorValue& descriptor, bool r128 = false) {
 	const auto type   = static_cast<Prospero::ImageType>((descriptor.dwords[3] >> 28u) & 0xfu);
 	const auto format = static_cast<Prospero::BufferFormat>((descriptor.dwords[1] >> 20u) & 0x1ffu);
-	if (type < Prospero::ImageType::kColor1D || format == Prospero::BufferFormat::kInvalid) {
+	if (type < Prospero::ImageType::kColor1D || format == Prospero::BufferFormat::kInvalid ||
+	    format > Prospero::BufferFormat::kBc7Srgb) {
 		return false;
 	}
 	if (r128 && type != Prospero::ImageType::kColor1D && type != Prospero::ImageType::kColor2D &&

src/graphics/shader/recompiler/ir/passes/ResourceTracking.cpp

@@ -139,7 +139,8 @@ public:
 		}
 		std::erase_if(m_program.dynamic_reads, [&](Value value) {
 			const auto* inst = value.Resolve().TryInstruction();
-			return std::ranges::any_of(m_indirect_images, [&](const IndirectImagePlan& plan) {
+			return std::any_of(m_indirect_images.begin(), m_indirect_images.end(),
+			                   [&](const IndirectImagePlan& plan) {
 				return std::ranges::find(plan.reads, inst) != plan.reads.end();
 			});
 		});
@@ -432,14 +433,16 @@ private:
 
 	const IndirectImagePlan* FindIndirectImage(const Inst& handle) const {
 		const auto found =
-		    std::ranges::find_if(m_indirect_images, [&](const IndirectImagePlan& plan) {
+		    std::find_if(m_indirect_images.begin(), m_indirect_images.end(),
+		                 [&](const IndirectImagePlan& plan) {
 			    return plan.handle == &handle;
 		    });
 		return found == m_indirect_images.end() ? nullptr : &*found;
 	}
 
 	bool IsIndirectPlanningMemory(uint32_t index) const {
-		return std::ranges::any_of(m_indirect_images, [&](const IndirectImagePlan& plan) {
+		return std::any_of(m_indirect_images.begin(), m_indirect_images.end(),
+		                   [&](const IndirectImagePlan& plan) {
 			return std::ranges::find(plan.memory, index) != plan.memory.end();
 		});
 	}

src/graphics/shader/recompiler/ir/passes/SsaRewrite.cpp

@@ -22,9 +22,6 @@ struct ScalarMaskTag {
 struct ExecTag {
 	auto operator<=>(const ExecTag&) const = default;
 };
-struct ExecMaskTag {
-	auto operator<=>(const ExecMaskTag&) const = default;
-};
 struct ExecLoTag {
 	auto operator<=>(const ExecLoTag&) const = default;
 };
@@ -34,9 +31,6 @@ struct ExecHiTag {
 struct VccTag {
 	auto operator<=>(const VccTag&) const = default;
 };
-struct VccMaskValidTag {
-	auto operator<=>(const VccMaskValidTag&) const = default;
-};
 struct VccLoTag {
 	auto operator<=>(const VccLoTag&) const = default;
 };
@@ -51,9 +45,9 @@ struct GotoVariable {
 	auto     operator<=>(const GotoVariable&) const = default;
 };
 
-using Variable = std::variant<ScalarReg, ThreadBitScalarReg, ScalarMaskTag, VectorReg, GotoVariable,
-                              SccTag, ExecTag, ExecMaskTag, ExecLoTag, ExecHiTag, VccTag,
-                              VccMaskValidTag, VccLoTag, VccHiTag, M0Tag>;
+using Variable =
+    std::variant<ScalarReg, ThreadBitScalarReg, ScalarMaskTag, VectorReg, GotoVariable, SccTag,
+                 ExecTag, ExecLoTag, ExecHiTag, VccTag, VccLoTag, VccHiTag, M0Tag>;
 using ValueMap = std::unordered_map<Block*, Value>;
 
 struct DefTable {
@@ -105,16 +99,12 @@ struct DefTable {
 	void         Set(Block* block, SccTag, Value value) { scc[block] = value; }
 	const Value& Get(Block* block, ExecTag) { return exec[block]; }
 	void         Set(Block* block, ExecTag, Value value) { exec[block] = value; }
-	const Value& Get(Block* block, ExecMaskTag) { return exec_mask_tag[block]; }
-	void         Set(Block* block, ExecMaskTag, Value value) { exec_mask_tag[block] = value; }
 	const Value& Get(Block* block, ExecLoTag) { return exec_lo[block]; }
 	void         Set(Block* block, ExecLoTag, Value value) { exec_lo[block] = value; }
 	const Value& Get(Block* block, ExecHiTag) { return exec_hi[block]; }
 	void         Set(Block* block, ExecHiTag, Value value) { exec_hi[block] = value; }
 	const Value& Get(Block* block, VccTag) { return vcc[block]; }
 	void         Set(Block* block, VccTag, Value value) { vcc[block] = value; }
-	const Value& Get(Block* block, VccMaskValidTag) { return vcc_mask_valid_tag[block]; }
-	void Set(Block* block, VccMaskValidTag, Value value) { vcc_mask_valid_tag[block] = value; }
 	const Value& Get(Block* block, VccLoTag) { return vcc_lo[block]; }
 	void         Set(Block* block, VccLoTag, Value value) { vcc_lo[block] = value; }
 	const Value& Get(Block* block, VccHiTag) { return vcc_hi[block]; }
@@ -124,11 +114,9 @@ struct DefTable {
 
 	ValueMap                               scc;
 	ValueMap                               exec;
-	ValueMap                               exec_mask_tag;
 	ValueMap                               exec_lo;
 	ValueMap                               exec_hi;
 	ValueMap                               vcc;
-	ValueMap                               vcc_mask_valid_tag;
 	ValueMap                               vcc_lo;
 	ValueMap                               vcc_hi;
 	ValueMap                               m0;
@@ -156,9 +144,6 @@ ValueOpcode UndefOpcode(SccTag) {
 ValueOpcode UndefOpcode(ExecTag) {
 	return ValueOpcode::UndefU1;
 }
-ValueOpcode UndefOpcode(ExecMaskTag) {
-	return ValueOpcode::UndefU1;
-}
 ValueOpcode UndefOpcode(ExecLoTag) {
 	return ValueOpcode::UndefU32;
 }
@@ -168,9 +153,6 @@ ValueOpcode UndefOpcode(ExecHiTag) {
 ValueOpcode UndefOpcode(VccTag) {
 	return ValueOpcode::UndefU1;
 }
-ValueOpcode UndefOpcode(VccMaskValidTag) {
-	return ValueOpcode::UndefU1;
-}
 ValueOpcode UndefOpcode(VccLoTag) {
 	return ValueOpcode::UndefU32;
 }
@@ -202,9 +184,6 @@ Value InitialValue(SccTag) {
 Value InitialValue(ExecTag) {
 	return Value(false);
 }
-Value InitialValue(ExecMaskTag) {
-	return Value(false);
-}
 Value InitialValue(ExecLoTag) {
 	return Value(0u);
 }
@@ -214,9 +193,6 @@ Value InitialValue(ExecHiTag) {
 Value InitialValue(VccTag) {
 	return Value(false);
 }
-Value InitialValue(VccMaskValidTag) {
-	return Value(false);
-}
 Value InitialValue(VccLoTag) {
 	return Value(0u);
 }
@@ -375,15 +351,9 @@ void VisitInstruction(Pass& pass, Block* block, Inst& inst) {
 			break;
 		case ValueOpcode::SetScc: pass.Write(SccTag {}, block, inst.Arg(0)); break;
 		case ValueOpcode::SetExec: pass.Write(ExecTag {}, block, inst.Arg(0)); break;
-		case ValueOpcode::SetExecMaskTag:
-			pass.Write(ExecMaskTag {}, block, inst.Arg(0));
-			break;
 		case ValueOpcode::SetExecLo: pass.Write(ExecLoTag {}, block, inst.Arg(0)); break;
 		case ValueOpcode::SetExecHi: pass.Write(ExecHiTag {}, block, inst.Arg(0)); break;
 		case ValueOpcode::SetVcc: pass.Write(VccTag {}, block, inst.Arg(0)); break;
-		case ValueOpcode::SetVccMaskValidTag:
-			pass.Write(VccMaskValidTag {}, block, inst.Arg(0));
-			break;
 		case ValueOpcode::SetVccLo: pass.Write(VccLoTag {}, block, inst.Arg(0)); break;
 		case ValueOpcode::SetVccHi: pass.Write(VccHiTag {}, block, inst.Arg(0)); break;
 		case ValueOpcode::SetM0: pass.Write(M0Tag {}, block, inst.Arg(0)); break;
@@ -405,15 +375,9 @@ void VisitInstruction(Pass& pass, Block* block, Inst& inst) {
 			break;
 		case ValueOpcode::GetScc: inst.ReplaceUsesWith(pass.Read(SccTag {}, block)); break;
 		case ValueOpcode::GetExec: inst.ReplaceUsesWith(pass.Read(ExecTag {}, block)); break;
-		case ValueOpcode::GetExecMaskTag:
-			inst.ReplaceUsesWith(pass.Read(ExecMaskTag {}, block));
-			break;
 		case ValueOpcode::GetExecLo: inst.ReplaceUsesWith(pass.Read(ExecLoTag {}, block)); break;
 		case ValueOpcode::GetExecHi: inst.ReplaceUsesWith(pass.Read(ExecHiTag {}, block)); break;
 		case ValueOpcode::GetVcc: inst.ReplaceUsesWith(pass.Read(VccTag {}, block)); break;
-		case ValueOpcode::GetVccMaskValidTag:
-			inst.ReplaceUsesWith(pass.Read(VccMaskValidTag {}, block));
-			break;
 		case ValueOpcode::GetVccLo: inst.ReplaceUsesWith(pass.Read(VccLoTag {}, block)); break;
 		case ValueOpcode::GetVccHi: inst.ReplaceUsesWith(pass.Read(VccHiTag {}, block)); break;
 		case ValueOpcode::GetM0: inst.ReplaceUsesWith(pass.Read(M0Tag {}, block)); break;

src/graphics/shader/shader.cpp

@@ -25,6 +25,7 @@
 
 #include <algorithm>
 #include <atomic>
+#include <bit>
 #include <cstdio>
 #include <filesystem>
 #include <fmt/format.h>
@@ -173,6 +174,11 @@ static const ShaderBinaryInfo* GetBinaryInfo(const uint32_t* code) {
 	return nullptr;
 }
 
+static uint64_t GetDeclaredShaderHash(uint64_t shader_addr) {
+	const auto* header = GetBinaryInfo(reinterpret_cast<const uint32_t*>(shader_addr));
+	return header != nullptr ? (static_cast<uint64_t>(header->hash1) << 32u) | header->hash0 : 0;
+}
+
 static ShaderParams GetShaderParams(uint64_t shader_addr, const char* label, uint64_t declared_hash,
 	                                std::span<const uint32_t> user_data,
 	                                const ShaderMappedData& data) {
@@ -210,7 +216,6 @@ static void vs_check(const HW::VertexShaderInfo& vs, const HW::ShaderRegisters&
 	};
 	const bool ps5_ngg_passthrough_triangle_path =
 	    vs.es_regs.data_addr != 0 && vs.gs_regs.data_addr == vs.es_regs.data_addr &&
-	    vs.gs_regs.chksum != 0 &&
 	    sh.m_geNggSubgrpCntl == 0x00000001 && sh.m_vgtGsMaxVertOut == 0x00000003 &&
 	    sh.m_vgtGsOutPrimType == 0x00000002 && sh.m_geMaxOutputPerSubgroup <= 0x000000c0;
 
@@ -275,7 +280,6 @@ static void ps_check(const HW::PsStageRegisters& ps, const HW::ShaderRegisters&
 	// EXIT_NOT_IMPLEMENTED(ps.m_spiShaderPgmRsrc1Ps != 0x002c0000);
 	// EXIT_NOT_IMPLEMENTED(ps.m_spiShaderPgmRsrc2Ps != 0x00000000);
 	// EXIT_NOT_IMPLEMENTED(ps.vgprs != 0x00 && ps.vgprs != 0x01);
-	// EXIT_NOT_IMPLEMENTED(ps.sgprs != 0x00 && ps.sgprs != 0x01);
 	EXIT_NOT_IMPLEMENTED(ps.rsrc1.priority != 0);
 	EXIT_NOT_IMPLEMENTED(ps.rsrc1.float_mode != 192);
 	EXIT_NOT_IMPLEMENTED(ps.rsrc1.dx10_clamp != true);
@@ -365,8 +369,10 @@ static void cs_check(const HW::CsStageRegisters& cs, const HW::ShaderRegisters&
 	// EXIT_NOT_IMPLEMENTED(cs.num_thread_y != 0x00000001);
 	// EXIT_NOT_IMPLEMENTED(cs.num_thread_z != 0x00000001);
 	// EXIT_NOT_IMPLEMENTED(cs.vgprs != 0x00 && cs.vgprs != 0x01);
-	// EXIT_NOT_IMPLEMENTED(cs.sgprs != 0x01 && cs.sgprs != 0x02);
-	EXIT_NOT_IMPLEMENTED(cs.bulky != 0x00);
+	EXIT_NOT_IMPLEMENTED(cs.priority != 0x00);
+	EXIT_NOT_IMPLEMENTED(cs.debug_mode != false);
+	EXIT_NOT_IMPLEMENTED(cs.require_forward_progress != false);
+	EXIT_NOT_IMPLEMENTED(cs.shared_vgprs != 0x00);
 	EXIT_NOT_IMPLEMENTED(cs.scratch_en != 0x00);
 	// EXIT_NOT_IMPLEMENTED(cs.user_sgpr != 0x0c);
 	if (cs.tgid_x_en == 0x00) {
@@ -566,7 +572,10 @@ static void ShaderApplyAttribSemantics(ShaderVertexInputInfo& info,
 		uint32_t reg  = in.hardware_mapping;
 		uint32_t size = in.size_in_elements;
 
-		LOGF("reg = %u, size = %u, va[%u] = 0x%08" PRIx32 "\n", reg, size, i, attrib[in.semantic]);
+		if (Config::GraphicsDebugDumpEnabled()) {
+			LOGF("reg = %u, size = %u, va[%u] = 0x%08" PRIx32 "\n", reg, size, i,
+			     attrib[in.semantic]);
+		}
 
 		size_t index = attrib[in.semantic] & 0x1fu;
 		auto   format =
@@ -689,7 +698,7 @@ static bool ShaderGetStaticInputInfoVS(const HW::VertexShaderInfo& regs,
 	info.export_count      = static_cast<int>(sh.GetExportCount());
 	info.pa_cl_vs_out_cntl = sh.m_paClVsOutCntl;
 
-	EXIT_NOT_IMPLEMENTED(regs.es_regs.data_addr == 0 || regs.gs_regs.chksum == 0);
+	EXIT_NOT_IMPLEMENTED(regs.es_regs.data_addr == 0);
 
 	uint64_t                shader_addr   = regs.es_regs.data_addr;
 	const HW::UserSgprInfo& user_sgpr     = regs.gs_user_sgpr;
@@ -698,8 +707,8 @@ static bool ShaderGetStaticInputInfoVS(const HW::VertexShaderInfo& regs,
 
 	if (data.user_data == nullptr) {
 		LOGF("ShaderGetInputInfoVS(): no AGC user data for shader=0x%016" PRIx64 " es=0x%016" PRIx64
-		     " gs=0x%016" PRIx64 " chksum=0x%016" PRIx64 " user_sgpr_num=%u\n",
-		     shader_addr, regs.es_regs.data_addr, regs.gs_regs.data_addr, regs.gs_regs.chksum,
+		     " gs=0x%016" PRIx64 " user_sgpr_num=%u\n",
+		     shader_addr, regs.es_regs.data_addr, regs.gs_regs.data_addr,
 		     static_cast<uint32_t>(user_sgpr_num));
 	}
 	ShaderVertexMetadata metadata;
@@ -889,6 +898,18 @@ void BuildStageStaticKey(const ShaderVertexInputInfo& info, std::vector<uint32_t
 	key.push_back(info.resources_num);
 	key.push_back(info.scratch_size_dwords);
 	key.push_back(info.pa_cl_vs_out_cntl);
+	key.push_back(static_cast<uint32_t>(info.clip_space.enabled));
+	if (info.clip_space.enabled) {
+		for (const float value: info.clip_space.scale) {
+			key.push_back(std::bit_cast<uint32_t>(value));
+		}
+		for (const float value: info.clip_space.offset) {
+			key.push_back(std::bit_cast<uint32_t>(value));
+		}
+		for (const float value: info.clip_space.half_extent) {
+			key.push_back(std::bit_cast<uint32_t>(value));
+		}
+	}
 
 	for (int i = 0; i < info.resources_num; i++) {
 		const auto& resource    = info.resources[i];
@@ -979,7 +1000,7 @@ ShaderParams PrepareProgram(const HW::VertexShaderInfo& regs, const HW::ShaderRe
 		EXIT("failed to prepare vertex shader program\n");
 	}
 	return GetShaderParams(
-	    regs.es_regs.data_addr, "ShaderRecompiler VS", regs.gs_regs.chksum,
+	    regs.es_regs.data_addr, "ShaderRecompiler VS", GetDeclaredShaderHash(regs.es_regs.data_addr),
 	    std::span<const uint32_t>(regs.gs_user_sgpr.value, regs.gs_regs.rsrc2.user_sgpr), data);
 }
 
@@ -991,7 +1012,7 @@ ShaderParams PrepareProgram(
 	const auto data = ShaderGetMappedData(regs.ps_regs.data_addr, "ShaderGetInputInfoPS():");
 	ShaderGetStaticInputInfoPS(regs, sh, vs_info, target_export_mapping, data, ps_info);
 	return GetShaderParams(
-	    regs.ps_regs.data_addr, "ShaderRecompiler PS", regs.ps_regs.chksum,
+	    regs.ps_regs.data_addr, "ShaderRecompiler PS", GetDeclaredShaderHash(regs.ps_regs.data_addr),
 	    std::span<const uint32_t>(regs.ps_user_sgpr.value, regs.ps_regs.rsrc2.user_sgpr), data);
 }
 
@@ -999,11 +1020,8 @@ ShaderParams PrepareProgram(const HW::ComputeShaderInfo& regs, const HW::ShaderR
                             ShaderComputeInputInfo& info) {
 	const auto data = ShaderGetMappedData(regs.cs_regs.data_addr, "ShaderGetInputInfoCS():");
 	ShaderGetStaticInputInfoCS(regs, sh, data, info);
-	const auto* header = GetBinaryInfo(reinterpret_cast<const uint32_t*>(regs.cs_regs.data_addr));
-	const auto declared_hash =
-	    header != nullptr ? (static_cast<uint64_t>(header->hash1) << 32u) | header->hash0 : 0;
 	return GetShaderParams(
-	    regs.cs_regs.data_addr, "ShaderRecompiler CS", declared_hash,
+	    regs.cs_regs.data_addr, "ShaderRecompiler CS", GetDeclaredShaderHash(regs.cs_regs.data_addr),
 	    std::span<const uint32_t>(regs.cs_user_sgpr.value, regs.cs_regs.user_sgpr), data);
 }
 
@@ -1234,9 +1252,7 @@ static ShaderRecompiler::CompileOptions MakeCompileOptions(const ShaderParams& p
 	ShaderRecompiler::CompileOptions options;
 	options.stage                      = stage;
 	options.shader_hash                = params.hash;
-	options.shader_base                = params.Base();
-	options.user_data_count            = static_cast<uint32_t>(params.user_data.size());
-	options.user_data                  = params.user_data.data();
+	options.user_data                  = params.user_data;
 	options.read_specialization_memory = ReadShaderGuestMemory;
 	options.dump_ir                    = ShaderRecompilerTextDumpEnabled();
 	options.early_dump                 = options.dump_ir;

src/graphics/shader/shader.h

@@ -68,6 +68,13 @@ inline uint8_t GetDstSel(uint32_t swizzle, uint32_t channel) {
 	return (swizzle >> (channel * 3u)) & 0x7u;
 }
 
+struct ShaderClipSpaceTransform {
+	float scale[2]       = {};
+	float offset[2]      = {};
+	float half_extent[2] = {};
+	bool  enabled        = false;
+};
+
 struct ShaderVertexInputInfo {
 	static constexpr int RES_MAX = 32;
 
@@ -84,6 +91,7 @@ struct ShaderVertexInputInfo {
 	uint32_t                scratch_size_dwords = 0;
 	uint32_t                param_export_mask   = 0;
 	uint32_t                pa_cl_vs_out_cntl    = 0;
+	ShaderClipSpaceTransform clip_space;
 	bool                    fetch_external      = false;
 	bool                    fetch_embedded      = false;
 };

src/launcher/src/inputMappingDialog.cpp

@@ -51,7 +51,8 @@ constexpr PadControl PAD_CONTROLS[] = {
     {"L3", "L3", "Left Shift"},
     {"R3", "R3", "Left Ctrl"},
     {"Options", "Options", "Return"},
-    {"TouchPad", "Touch pad", "Backspace"},
+    {"TouchPad", "Touch pad left (SELECT)", "Backspace"},
+    {"TouchPadRight", "Touch pad right (START)", "Tab"},
 };
 
 QString KeypadName(int key) {

src/libs/agc.cpp

@@ -44,6 +44,13 @@ namespace Libs::Graphics {
 
 static RenderContext* g_renderer = nullptr;
 
+template <typename... Args>
+static void AgcTrace(const char* format, const Args&... args) {
+	if (Config::GraphicsDebugDumpEnabled()) {
+		LOGF(format, args...);
+	}
+}
+
 void Initialize() {
 	// Some games lock up if this is not called first
 	if (Config::RenderDocEnabled()) {
@@ -145,7 +152,6 @@ struct RegisterDefaultsStorage {
 
 static constexpr uint32_t GRAPHICS_REGISTER_DEFAULTS_MAX_VERSION      = 13;
 static constexpr uint32_t GRAPHICS_REGISTER_DEFAULTS_FALLBACK_VERSION = 13;
-static constexpr uint32_t GRAPHICS_INIT_NO_FEATURE_STATE              = 0;
 
 static std::mutex g_register_defaults_mutex;
 
@@ -336,20 +342,12 @@ int KYTY_SYSV_ABI AgcInit(uint32_t* state, uint32_t ver) {
 	     "\t ver   = %u\n",
 	     reinterpret_cast<uint64_t>(state), ver);
 
-	if (state == nullptr) {
-		LOGF_COLOR(Log::Color::Red, "\t state is null\n");
-		return OK;
-	}
-
 	if (ver > GRAPHICS_REGISTER_DEFAULTS_MAX_VERSION) {
 		LOGF_COLOR(Log::Color::Red, "\t unsupported version %u\n", ver);
 	}
 
 	printf("version = %u\n", ver);
 
-	state[0] = ver;
-	state[1] = GRAPHICS_INIT_NO_FEATURE_STATE;
-
 	return OK;
 }
 
@@ -372,6 +370,10 @@ void* KYTY_SYSV_ABI AgcGetRegisterDefaults2Internal(uint32_t ver) {
 }
 
 static void dbg_dump_shader(const Shader* h) {
+	if (!Config::GraphicsDebugDumpEnabled()) {
+		return;
+	}
+
 	LOGF("\t file_header  = 0x%08" PRIx32 "\n"
 	     "\t version      = 0x%08" PRIx32 "\n"
 	     "\t user_data    = 0x%016" PRIx64 "\n",
@@ -903,8 +905,12 @@ static bool is_native_reg_indirect_packet(const uint32_t* cmd, RegIndirectPacket
 	return ((cmd[0] >> 8u) & 0xffu) == reg_indirect_native_op(type);
 }
 
+static constexpr bool is_trinity_mode() {
+	return false;
+}
+
 static uint32_t reg_indirect_pm4_r(RegIndirectPacket type) {
-	return (type == RegIndirectPacket::Uc && AgcGetIsTrinityMode() != 0 ? 1u : 0u);
+	return (type == RegIndirectPacket::Uc && is_trinity_mode() ? 1u : 0u);
 }
 
 static void reg_indirect_write_packet(uint32_t* cmd, uint64_t vaddr, uint32_t num_regs,
@@ -957,7 +963,7 @@ int KYTY_SYSV_ABI AgcSetCxRegIndirectPatchSetAddress(uint32_t*
                                                      const volatile ShaderRegister* regs) {
 	PRINT_NAME();
 
-	LOGF("\t cmd  = 0x%016" PRIx64 "\n"
+	AgcTrace("\t cmd  = 0x%016" PRIx64 "\n"
 	     "\t regs = 0x%016" PRIx64 "\n",
 	     reinterpret_cast<uint64_t>(cmd), reinterpret_cast<uint64_t>(regs));
 
@@ -973,7 +979,7 @@ int KYTY_SYSV_ABI AgcSetShRegIndirectPatchSetAddress(uint32_t*
                                                      const volatile ShaderRegister* regs) {
 	PRINT_NAME();
 
-	LOGF("\t cmd  = 0x%016" PRIx64 "\n"
+	AgcTrace("\t cmd  = 0x%016" PRIx64 "\n"
 	     "\t regs = 0x%016" PRIx64 "\n",
 	     reinterpret_cast<uint64_t>(cmd), reinterpret_cast<uint64_t>(regs));
 
@@ -989,7 +995,7 @@ int KYTY_SYSV_ABI AgcSetUcRegIndirectPatchSetAddress(uint32_t*
                                                      const volatile ShaderRegister* regs) {
 	PRINT_NAME();
 
-	LOGF("\t cmd  = 0x%016" PRIx64 "\n"
+	AgcTrace("\t cmd  = 0x%016" PRIx64 "\n"
 	     "\t regs = 0x%016" PRIx64 "\n",
 	     reinterpret_cast<uint64_t>(cmd), reinterpret_cast<uint64_t>(regs));
 
@@ -1004,7 +1010,7 @@ int KYTY_SYSV_ABI AgcSetUcRegIndirectPatchSetAddress(uint32_t*
 int KYTY_SYSV_ABI AgcSetCxRegIndirectPatchSetNumRegisters(uint32_t* cmd, uint32_t num_regs) {
 	PRINT_NAME();
 
-	LOGF("\t cmd      = 0x%016" PRIx64 "\n"
+	AgcTrace("\t cmd      = 0x%016" PRIx64 "\n"
 	     "\t num_regs = %" PRIu32 "\n",
 	     reinterpret_cast<uint64_t>(cmd), num_regs);
 
@@ -1016,7 +1022,7 @@ int KYTY_SYSV_ABI AgcSetCxRegIndirectPatchSetNumRegisters(uint32_t* cmd, uint32_
 int KYTY_SYSV_ABI AgcSetShRegIndirectPatchSetNumRegisters(uint32_t* cmd, uint32_t num_regs) {
 	PRINT_NAME();
 
-	LOGF("\t cmd      = 0x%016" PRIx64 "\n"
+	AgcTrace("\t cmd      = 0x%016" PRIx64 "\n"
 	     "\t num_regs = %" PRIu32 "\n",
 	     reinterpret_cast<uint64_t>(cmd), num_regs);
 
@@ -1028,7 +1034,7 @@ int KYTY_SYSV_ABI AgcSetShRegIndirectPatchSetNumRegisters(uint32_t* cmd, uint32_
 int KYTY_SYSV_ABI AgcSetUcRegIndirectPatchSetNumRegisters(uint32_t* cmd, uint32_t num_regs) {
 	PRINT_NAME();
 
-	LOGF("\t cmd      = 0x%016" PRIx64 "\n"
+	AgcTrace("\t cmd      = 0x%016" PRIx64 "\n"
 	     "\t num_regs = %" PRIu32 "\n",
 	     reinterpret_cast<uint64_t>(cmd), num_regs);
 
@@ -1040,7 +1046,7 @@ int KYTY_SYSV_ABI AgcSetUcRegIndirectPatchSetNumRegisters(uint32_t* cmd, uint32_
 int KYTY_SYSV_ABI AgcSetCxRegIndirectPatchAddRegisters(uint32_t* cmd, uint32_t num_regs) {
 	PRINT_NAME();
 
-	LOGF("\t cmd      = 0x%016" PRIx64 "\n"
+	AgcTrace("\t cmd      = 0x%016" PRIx64 "\n"
 	     "\t num_regs = %" PRIu32 "\n",
 	     reinterpret_cast<uint64_t>(cmd), num_regs);
 
@@ -1052,7 +1058,7 @@ int KYTY_SYSV_ABI AgcSetCxRegIndirectPatchAddRegisters(uint32_t* cmd, uint32_t n
 int KYTY_SYSV_ABI AgcSetShRegIndirectPatchAddRegisters(uint32_t* cmd, uint32_t num_regs) {
 	PRINT_NAME();
 
-	LOGF("\t cmd      = 0x%016" PRIx64 "\n"
+	AgcTrace("\t cmd      = 0x%016" PRIx64 "\n"
 	     "\t num_regs = %" PRIu32 "\n",
 	     reinterpret_cast<uint64_t>(cmd), num_regs);
 
@@ -1064,7 +1070,7 @@ int KYTY_SYSV_ABI AgcSetShRegIndirectPatchAddRegisters(uint32_t* cmd, uint32_t n
 int KYTY_SYSV_ABI AgcSetUcRegIndirectPatchAddRegisters(uint32_t* cmd, uint32_t num_regs) {
 	PRINT_NAME();
 
-	LOGF("\t cmd      = 0x%016" PRIx64 "\n"
+	AgcTrace("\t cmd      = 0x%016" PRIx64 "\n"
 	     "\t num_regs = %" PRIu32 "\n",
 	     reinterpret_cast<uint64_t>(cmd), num_regs);
 
@@ -1095,7 +1101,7 @@ int KYTY_SYSV_ABI AgcCreatePrimState(ShaderRegister* cx_regs, ShaderRegister* uc
                                      const Shader* hs, const Shader* gs, uint32_t prim_type) {
 	PRINT_NAME();
 
-	LOGF("\t cx_regs   = 0x%016" PRIx64 "\n"
+	AgcTrace("\t cx_regs   = 0x%016" PRIx64 "\n"
 	     "\t uc_regs   = 0x%016" PRIx64 "\n"
 	     "\t hs        = 0x%016" PRIx64 "\n"
 	     "\t gs        = 0x%016" PRIx64 "\n"
@@ -1163,7 +1169,7 @@ int KYTY_SYSV_ABI AgcUpdatePrimState(ShaderRegister* cx_regs, ShaderRegister* uc
                                      uint32_t prim_type) {
 	PRINT_NAME();
 
-	LOGF("\t cx_regs   = 0x%016" PRIx64 "\n"
+	AgcTrace("\t cx_regs   = 0x%016" PRIx64 "\n"
 	     "\t uc_regs   = 0x%016" PRIx64 "\n"
 	     "\t prim_type = %" PRIu32 "\n",
 	     reinterpret_cast<uint64_t>(cx_regs), reinterpret_cast<uint64_t>(uc_regs), prim_type);
@@ -1225,7 +1231,7 @@ int KYTY_SYSV_ABI AgcGetGsOversubscription(ShaderRegister* regs, const Shader* g
 	constexpr uint32_t FULL_PC_OVERSUBSCRIPTION = 0x7ffu;
 	constexpr uint32_t FULL_SH_OVERSUBSCRIPTION = 0x007f0000u;
 
-	regs[0] = {Pm4::GE_PC_ALLOC, 0u};
+	regs[0] = {Pm4::UC_PARAMETER_OVERSUBSCRIPTION, 0u};
 	regs[1] = {Pm4::SPI_SHADER_PGM_RSRC4_GS, 0u};
 
 	if (budget == 0u) {
@@ -1368,7 +1374,7 @@ int KYTY_SYSV_ABI AgcCreateInterpolantMapping(ShaderRegister* regs, const Shader
                                               const Shader* ps) {
 	PRINT_NAME();
 
-	LOGF("\t regs = 0x%016" PRIx64 "\n"
+	AgcTrace("\t regs = 0x%016" PRIx64 "\n"
 	     "\t gs   = 0x%016" PRIx64 "\n"
 	     "\t ps   = 0x%016" PRIx64 "\n",
 	     reinterpret_cast<uint64_t>(regs), reinterpret_cast<uint64_t>(gs),
@@ -1424,7 +1430,7 @@ int KYTY_SYSV_ABI AgcCreateInterpolantMapping2(ShaderRegister* regs, const Shade
                                                const Shader* ps) {
 	PRINT_NAME();
 
-	LOGF("\t regs = 0x%016" PRIx64 "\n"
+	AgcTrace("\t regs = 0x%016" PRIx64 "\n"
 	     "\t gs   = 0x%016" PRIx64 "\n"
 	     "\t ps   = 0x%016" PRIx64 "\n",
 	     reinterpret_cast<uint64_t>(regs), reinterpret_cast<uint64_t>(gs),
@@ -1574,15 +1580,6 @@ int KYTY_SYSV_ABI AgcWriteDataPatchSetAddressOrOffset(uint32_t* cmd, uint64_t ad
 	return GRAPHICS5_ERROR_INVALID_PACKET;
 }
 
-int KYTY_SYSV_ABI AgcUnknownJumpPatchSetTarget(uint32_t* cmd, const volatile uint32_t* target,
-                                               uint32_t size_in_dwords) {
-	PRINT_NAME();
-
-	LOGF("Calling unknown stub");
-
-	return AgcJumpPatchSetTarget(cmd, target, size_in_dwords);
-}
-
 int KYTY_SYSV_ABI AgcJumpPatchSetTarget(uint32_t* cmd, const volatile uint32_t* target,
                                         uint32_t size_in_dwords) {
 	PRINT_NAME();
@@ -1690,9 +1687,8 @@ uint64_t KYTY_SYSV_ABI AgcDcbContextStateOpGetSize(uint32_t operation) {
 	return static_cast<uint64_t>(context_state_op_size_dw(operation)) * sizeof(uint32_t);
 }
 
-uint64_t KYTY_SYSV_ABI AgcGetIsTrinityMode() {
-	// PS5 Pro?
-	return 0;
+void KYTY_SYSV_ABI AgcGetIsTrinityMode(uint8_t* result) {
+	*result = is_trinity_mode();
 }
 
 static constexpr int      GRAPHICS5_DRIVER_ERROR_INVALID_VALUE    = static_cast<int>(0x8a6c0033u);
@@ -1957,7 +1953,7 @@ uint32_t* KYTY_SYSV_ABI AgcCbSetShRegistersDirect(CommandBuffer*
                                                   uint32_t                       num_regs) {
 	PRINT_NAME();
 
-	LOGF("\t regs     = 0x%016" PRIx64 "\n"
+	AgcTrace("\t regs     = 0x%016" PRIx64 "\n"
 	     "\t num_regs = %" PRIu32 "\n",
 	     reinterpret_cast<uint64_t>(regs), num_regs);
 
@@ -2024,7 +2020,7 @@ uint32_t* KYTY_SYSV_ABI AgcCbSetUcRegistersDirect(CommandBuffer*
                                                   uint32_t                       num_regs) {
 	PRINT_NAME();
 
-	LOGF("\t regs     = 0x%016" PRIx64 "\n"
+	AgcTrace("\t regs     = 0x%016" PRIx64 "\n"
 	     "\t num_regs = %" PRIu32 "\n",
 	     reinterpret_cast<uint64_t>(regs), num_regs);
 
@@ -2411,7 +2407,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbSetCxRegistersIndirect(CommandBuffer*
                                                      uint32_t                       num_regs) {
 	PRINT_NAME();
 
-	LOGF("\t regs     = 0x%016" PRIx64 "\n"
+	AgcTrace("\t regs     = 0x%016" PRIx64 "\n"
 	     "\t num_regs = %" PRIu32 "\n",
 	     reinterpret_cast<uint64_t>(regs), num_regs);
 
@@ -2435,7 +2431,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbSetShRegistersIndirect(CommandBuffer*
                                                      uint32_t                       num_regs) {
 	PRINT_NAME();
 
-	LOGF("\t regs     = 0x%016" PRIx64 "\n"
+	AgcTrace("\t regs     = 0x%016" PRIx64 "\n"
 	     "\t num_regs = %" PRIu32 "\n",
 	     reinterpret_cast<uint64_t>(regs), num_regs);
 
@@ -2459,7 +2455,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbSetUcRegistersIndirect(CommandBuffer*
                                                      uint32_t                       num_regs) {
 	PRINT_NAME();
 
-	LOGF("\t regs     = 0x%016" PRIx64 "\n"
+	AgcTrace("\t regs     = 0x%016" PRIx64 "\n"
 	     "\t num_regs = %" PRIu32 "\n",
 	     reinterpret_cast<uint64_t>(regs), num_regs);
 
@@ -2482,7 +2478,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbSetIndexSize(CommandBuffer* buf, uint8_t index_siz
                                            uint8_t cache_policy) {
 	PRINT_NAME();
 
-	LOGF("\t index_size   = 0x%" PRIx8 "\n"
+	AgcTrace("\t index_size   = 0x%" PRIx8 "\n"
 	     "\t cache_policy = 0x%" PRIx8 "\n",
 	     index_size, cache_policy);
 
@@ -2503,7 +2499,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbSetIndexSize(CommandBuffer* buf, uint8_t index_siz
 uint32_t* KYTY_SYSV_ABI AgcDcbSetIndexBuffer(CommandBuffer* buf, uint64_t index_addr) {
 	PRINT_NAME();
 
-	LOGF("\t index_addr = 0x%016" PRIx64 "\n", index_addr);
+	AgcTrace("\t index_addr = 0x%016" PRIx64 "\n", index_addr);
 
 	EXIT_NOT_IMPLEMENTED(buf == nullptr);
 	EXIT_NOT_IMPLEMENTED((index_addr & 1u) != 0);
@@ -2524,7 +2520,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbSetIndexBuffer(CommandBuffer* buf, uint64_t index_
 uint32_t* KYTY_SYSV_ABI AgcDcbSetIndexCount(CommandBuffer* buf, uint32_t index_count) {
 	PRINT_NAME();
 
-	LOGF("\t index_count = 0x%" PRIx32 "\n", index_count);
+	AgcTrace("\t index_count = 0x%" PRIx32 "\n", index_count);
 
 	EXIT_NOT_IMPLEMENTED(buf == nullptr);
 
@@ -2543,7 +2539,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbSetIndexCount(CommandBuffer* buf, uint32_t index_c
 uint32_t* KYTY_SYSV_ABI AgcDcbSetNumInstances(CommandBuffer* buf, uint32_t num_instances) {
 	PRINT_NAME();
 
-	LOGF("\t num_instances = 0x%" PRIx32 "\n", num_instances);
+	AgcTrace("\t num_instances = 0x%" PRIx32 "\n", num_instances);
 
 	if (buf == nullptr) {
 		return nullptr;
@@ -2639,7 +2635,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbDrawIndex(CommandBuffer* buf, uint32_t index_count
                                         const volatile void* index_addr, uint64_t modifier) {
 	PRINT_NAME();
 
-	LOGF("\t index_count = 0x%" PRIx32 "\n"
+	AgcTrace("\t index_count = 0x%" PRIx32 "\n"
 	     "\t index_addr  = 0x%016" PRIx64 "\n"
 	     "\t modifier    = 0x%016" PRIx64 "\n",
 	     index_count, reinterpret_cast<uint64_t>(index_addr), modifier);
@@ -2678,7 +2674,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbDrawIndexMultiInstanced(CommandBuffer* buf, uint32
                                                       uint32_t instance_count, uint64_t modifier) {
 	PRINT_NAME();
 
-	LOGF("\t index_count    = 0x%" PRIx32 "\n"
+	AgcTrace("\t index_count    = 0x%" PRIx32 "\n"
 	     "\t index_addr     = 0x%016" PRIx64 "\n"
 	     "\t instance_count = 0x%" PRIx32 "\n"
 	     "\t object_ids     = 0x%016" PRIx64 "\n"
@@ -2754,7 +2750,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbDrawIndexAuto(CommandBuffer* buf, uint32_t index_c
                                             uint64_t modifier) {
 	PRINT_NAME();
 
-	LOGF("\t index_count = 0x%" PRIx32 "\n"
+	AgcTrace("\t index_count = 0x%" PRIx32 "\n"
 	     "\t modifier    = 0x%016" PRIx64 "\n",
 	     index_count, modifier);
 
@@ -2783,7 +2779,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbDrawIndexOffset(CommandBuffer* buf, uint32_t index
                                               uint32_t index_count, uint64_t modifier) {
 	PRINT_NAME();
 
-	LOGF("\t index_offset = 0x%" PRIx32 "\n"
+	AgcTrace("\t index_offset = 0x%" PRIx32 "\n"
 	     "\t index_count  = 0x%" PRIx32 "\n"
 	     "\t modifier     = 0x%016" PRIx64 "\n",
 	     index_offset, index_count, modifier);
@@ -2841,7 +2837,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbDrawIndexIndirect(CommandBuffer* buf, uint32_t dat
                                                 uint64_t modifier) {
 	PRINT_NAME();
 
-	LOGF("\t data_offset = 0x%" PRIx32 "\n"
+	AgcTrace("\t data_offset = 0x%" PRIx32 "\n"
 	     "\t modifier    = 0x%016" PRIx64 "\n",
 	     data_offset_in_bytes, modifier);
 
@@ -2868,7 +2864,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbDrawIndirect(CommandBuffer* buf, uint32_t data_off
                                            uint64_t modifier) {
 	PRINT_NAME();
 
-	LOGF("\t data_offset = 0x%" PRIx32 "\n"
+	AgcTrace("\t data_offset = 0x%" PRIx32 "\n"
 	     "\t modifier    = 0x%016" PRIx64 "\n",
 	     data_offset_in_bytes, modifier);
 
@@ -2899,7 +2895,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbDrawIndirectMulti(CommandBuffer*       buf,
                                                 uint32_t stride_in_bytes, uint64_t modifier) {
 	PRINT_NAME();
 
-	LOGF("\t data_offset        = 0x%" PRIx32 "\n"
+	AgcTrace("\t data_offset        = 0x%" PRIx32 "\n"
 	     "\t count_indirect     = 0x%" PRIx32 "\n"
 	     "\t max_count_or_count = 0x%" PRIx32 "\n"
 	     "\t count_addr         = 0x%016" PRIx64 "\n"
@@ -2962,7 +2958,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbDrawIndexIndirectMulti(CommandBuffer*       buf,
                                                      uint32_t stride_in_bytes, uint64_t modifier) {
 	PRINT_NAME();
 
-	LOGF("\t data_offset        = 0x%" PRIx32 "\n"
+	AgcTrace("\t data_offset        = 0x%" PRIx32 "\n"
 	     "\t count_indirect     = 0x%" PRIx32 "\n"
 	     "\t max_count_or_count = 0x%" PRIx32 "\n"
 	     "\t count_addr         = 0x%016" PRIx64 "\n"
@@ -3816,7 +3812,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbWriteData(CommandBuffer* buf, uint8_t dst, uint8_t
                                         uint8_t write_confirm) {
 	PRINT_NAME();
 
-	LOGF("\t dst               = 0x%02" PRIx8 "\n"
+	AgcTrace("\t dst               = 0x%02" PRIx8 "\n"
 	     "\t cache_policy      = 0x%02" PRIx8 "\n"
 	     "\t address_or_offset = 0x%016" PRIx64 "\n"
 	     "\t data              = 0x%016" PRIx64 "\n"
@@ -4069,7 +4065,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbWaitRegMem(CommandBuffer* buf, uint8_t size, uint8
                                          uint64_t mask, uint32_t poll_cycles) {
 	PRINT_NAME();
 
-	LOGF("\t size             = 0x%02" PRIx8 "\n"
+	AgcTrace("\t size             = 0x%02" PRIx8 "\n"
 	     "\t compare_function = 0x%02" PRIx8 "\n"
 	     "\t op               = 0x%02" PRIx8 "\n"
 	     "\t cache_policy     = 0x%02" PRIx8 "\n"
@@ -4216,7 +4212,7 @@ uint32_t* KYTY_SYSV_ABI AgcDcbSetFlip(CommandBuffer* buf, uint32_t video_out_han
                                       int64_t flip_arg) {
 	PRINT_NAME();
 
-	LOGF("\t video_out_handle     = %" PRIu32 "\n"
+	AgcTrace("\t video_out_handle     = %" PRIu32 "\n"
 	     "\t display_buffer_index = %" PRId32 "\n"
 	     "\t flip_mode            = %" PRIu32 "\n"
 	     "\t flip_arg             = %" PRId64 "\n",
@@ -4267,7 +4263,7 @@ int KYTY_SYSV_ABI AgcDriverSubmitDcb(const Packet* packet) {
 
 	EXIT_NOT_IMPLEMENTED(packet == nullptr);
 
-	LOGF("\t addr   = 0x%016" PRIx64 "\n"
+	AgcTrace("\t addr   = 0x%016" PRIx64 "\n"
 	     "\t dw_num = 0x%08" PRIx32 "\n"
 	     "\t flags  = 0x%02" PRIx8 "\n",
 	     reinterpret_cast<uint64_t>(packet->addr), packet->dw_num, packet->flags);
@@ -4281,7 +4277,7 @@ int KYTY_SYSV_ABI AgcDriverSubmitMultiDcbs(uint32_t* const* dcb_gpu_addrs,
                                            const uint32_t* dcb_sizes_in_dwords, uint32_t count) {
 	PRINT_NAME();
 
-	LOGF("\t count = %" PRIu32 "\n", count);
+	AgcTrace("\t count = %" PRIu32 "\n", count);
 
 	if (count == 0) {
 		return OK;
@@ -4294,7 +4290,7 @@ int KYTY_SYSV_ABI AgcDriverSubmitMultiDcbs(uint32_t* const* dcb_gpu_addrs,
 		auto*    dcb            = dcb_gpu_addrs[i];
 		uint32_t size_in_dwords = dcb_sizes_in_dwords[i];
 
-		LOGF("\t dcb[%" PRIu32 "]  = 0x%016" PRIx64 "\n"
+		AgcTrace("\t dcb[%" PRIu32 "]  = 0x%016" PRIx64 "\n"
 		     "\t size[%" PRIu32 "] = 0x%08" PRIx32 "\n",
 		     i, reinterpret_cast<uint64_t>(dcb), i, size_in_dwords);
 

src/libs/agc.h

@@ -71,14 +71,12 @@ int KYTY_SYSV_ABI AgcCreateInterpolantMapping2(ShaderRegister* regs, const Shade
 int KYTY_SYSV_ABI AgcGetDataPacketPayloadAddress(uint32_t** addr, uint32_t* cmd, int type);
 int KYTY_SYSV_ABI AgcGetDataPacketPayloadRange(MemoryRange* range, uint32_t* cmd, int type);
 int KYTY_SYSV_ABI AgcWriteDataPatchSetAddressOrOffset(uint32_t* cmd, uint64_t address_or_offset);
-int KYTY_SYSV_ABI AgcUnknownJumpPatchSetTarget(uint32_t* cmd, const volatile uint32_t* target,
-                                               uint32_t size_in_dwords);
 int KYTY_SYSV_ABI AgcJumpPatchSetTarget(uint32_t* cmd, const volatile uint32_t* target,
                                         uint32_t size_in_dwords);
 int KYTY_SYSV_ABI AgcSuspendPoint();
 uint32_t* KYTY_SYSV_ABI AgcDcbContextStateOp(CommandBuffer* buf, uint32_t operation);
 uint64_t KYTY_SYSV_ABI  AgcDcbContextStateOpGetSize(uint32_t operation);
-uint64_t KYTY_SYSV_ABI  AgcGetIsTrinityMode();
+void KYTY_SYSV_ABI      AgcGetIsTrinityMode(uint8_t* result);
 
 uint32_t KYTY_SYSV_ABI AgcDriverGetDefaultOwner();
 uint32_t KYTY_SYSV_ABI AgcDriverGetResourceRegistrationMaxNameLength();

src/libs/audio.cpp

@@ -2931,7 +2931,6 @@ int KYTY_SYSV_ABI Ngs2VoiceGetState(uintptr_t voice_handle, Ngs2VoiceState* stat
 			auto* submixer                    = reinterpret_cast<Ngs2SubmixerVoiceState*>(state);
 			*submixer                         = {};
 			submixer->voice_state.state_flags = Ngs2GetStateFlags(voice);
-			LOGF("\t state_flags = %u\n", submixer->voice_state.state_flags);
 			break;
 		}
 		case Ngs2RackType::CustomMastering: {
@@ -2942,7 +2941,6 @@ int KYTY_SYSV_ABI Ngs2VoiceGetState(uintptr_t voice_handle, Ngs2VoiceState* stat
 			std::memset(state, 0, state_size);
 			auto* mastering = reinterpret_cast<Ngs2CustomMasteringVoiceState*>(state);
 			mastering->voice_state.state_flags = Ngs2GetStateFlags(voice);
-			LOGF("\t state_flags = %u\n", mastering->voice_state.state_flags);
 			break;
 		}
 		case Ngs2RackType::Sampler:
@@ -2957,7 +2955,6 @@ int KYTY_SYSV_ABI Ngs2VoiceGetState(uintptr_t voice_handle, Ngs2VoiceState* stat
 
 			state->state_flags = Ngs2GetStateFlags(voice);
 			if (state_size < sizeof(Ngs2SamplerVoiceState)) {
-				LOGF("\t state_flags = %u\n", state->state_flags);
 				break;
 			}
 
@@ -2968,7 +2965,6 @@ int KYTY_SYSV_ABI Ngs2VoiceGetState(uintptr_t voice_handle, Ngs2VoiceState* stat
 			sampler->num_decoded_samples = 0;
 			sampler->user_data           = 0;
 			sampler->waveform_data       = nullptr;
-			LOGF("\t state_flags = %u\n", sampler->voice_state.state_flags);
 			break;
 		}
 		default: EXIT("unknown type: %s\n", Common::EnumName(voice->rack->type).c_str());
@@ -2989,8 +2985,6 @@ int KYTY_SYSV_ABI Ngs2VoiceGetStateFlags(uintptr_t voice_handle, uint32_t* state
 
 	*state_flags = Ngs2GetStateFlags(voice);
 
-	LOGF("\t state_flags = %u\n", *state_flags);
-
 	return OK;
 }
 

src/libs/controller.cpp

@@ -45,15 +45,22 @@ struct PadVibrationParam {
 };
 
 struct ControllerState {
+	struct Touch {
+		uint8_t  id   = 0;
+		bool     down = false;
+		uint16_t x    = 0;
+		uint16_t y    = 0;
+	};
+
 	uint64_t time                                  = 0;
 	uint32_t buttons                               = 0;
 	int      axes[static_cast<int>(Axis::AxisMax)] = {128, 128, 128, 128, 0, 0};
+	Touch    touch[2];
 };
 
 class GameController {
 public:
-	GameController()          = default;
-	virtual ~GameController() = default;
+	GameController() = default;
 
 	KYTY_CLASS_NO_COPY(GameController);
 
@@ -62,6 +69,7 @@ public:
 	void Button(int id, uint32_t button, bool down);
 	void Axis(int id, Axis axis, int value);
 	void RightStick(int id, int x, int y);
+	void TouchPad(int id, int finger, bool down, float x, float y);
 	void ResetInputState();
 	void GetConnectionInfo(bool* flag, int* count);
 	void SetVibration(uint8_t large_motor, uint8_t small_motor);
@@ -71,51 +79,49 @@ public:
 private:
 	static constexpr uint32_t STATES_MAX = 64;
 
-	struct StatePrivate {
-		bool obtained = false;
-	};
-
-	void                          CheckActive();
-	[[nodiscard]] ControllerState GetLastState() const;
-	void                          AddState(const ControllerState& state);
+	void CheckActive();
+	void AddState();
 
 	Common::Mutex    m_mutex;
 	std::vector<int> m_connected_ids;
 	int              m_active_id       = -1;
 	bool             m_connected       = false;
 	int              m_connected_count = 0;
+	ControllerState  m_state;
 	ControllerState  m_states[STATES_MAX];
-	StatePrivate     m_private[STATES_MAX];
-	ControllerState  m_last_state;
-	uint32_t         m_states_num  = 0;
-	uint32_t         m_first_state = 0;
+	bool             m_obtained[STATES_MAX] {};
+	uint32_t         m_states_num    = 0;
+	uint32_t         m_first_state   = 0;
+	uint8_t          m_next_touch_id = 1;
 };
 
 static GameController* g_controller = nullptr;
 
-static uint8_t pad_connected_count_to_u8(int connected_count) {
-	return static_cast<uint8_t>(connected_count > 255 ? 255 : connected_count);
-}
-
 static void pad_fill_data(PadData* data, const ControllerState& state, bool connected,
                           int connected_count) {
 	EXIT_IF(data == nullptr);
 
 	std::memset(data, 0, sizeof(*data));
 
-	data->buttons                = state.buttons;
-	data->left_stick_x           = state.axes[static_cast<int>(Axis::LeftX)];
-	data->left_stick_y           = state.axes[static_cast<int>(Axis::LeftY)];
-	data->right_stick_x          = state.axes[static_cast<int>(Axis::RightX)];
-	data->right_stick_y          = state.axes[static_cast<int>(Axis::RightY)];
-	data->analog_buttons_l2      = state.axes[static_cast<int>(Axis::TriggerLeft)];
-	data->analog_buttons_r2      = state.axes[static_cast<int>(Axis::TriggerRight)];
-	data->orientation_w          = 1.0f;
-	data->touch_data_touch0_id   = 1;
-	data->touch_data_touch1_id   = 2;
+	data->buttons           = state.buttons;
+	data->left_stick_x      = state.axes[static_cast<int>(Axis::LeftX)];
+	data->left_stick_y      = state.axes[static_cast<int>(Axis::LeftY)];
+	data->right_stick_x     = state.axes[static_cast<int>(Axis::RightX)];
+	data->right_stick_y     = state.axes[static_cast<int>(Axis::RightY)];
+	data->analog_buttons_l2 = state.axes[static_cast<int>(Axis::TriggerLeft)];
+	data->analog_buttons_r2 = state.axes[static_cast<int>(Axis::TriggerRight)];
+	data->orientation_w     = 1.0f;
+	for (const auto& touch: state.touch) {
+		if (touch.down) {
+			auto& output = data->touch_data.touch[data->touch_data.touch_num++];
+			output.x     = touch.x;
+			output.y     = touch.y;
+			output.id    = touch.id;
+		}
+	}
 	data->connected              = connected;
 	data->timestamp              = state.time;
-	data->connected_count        = pad_connected_count_to_u8(connected_count);
+	data->connected_count        = static_cast<uint8_t>(std::min(connected_count, 255));
 	data->device_unique_data_len = 0;
 }
 
@@ -155,7 +161,6 @@ void GameController::Disconnect(int id) {
 }
 
 void GameController::CheckActive() {
-	bool reset         = false;
 	int  new_active_id = -1;
 	bool new_connected = false;
 
@@ -170,46 +175,29 @@ void GameController::CheckActive() {
 		new_connected = true;
 	}
 
-	const bool was_connected = m_connected;
-
-	if (m_connected != new_connected || m_active_id != new_active_id) {
-		m_active_id = new_active_id;
-		m_connected = new_connected;
-		if (!was_connected && new_connected) {
-			m_connected_count++;
-		}
-		reset = true;
+	if (m_connected == new_connected && m_active_id == new_active_id) {
+		return;
 	}
-
-	if (reset) {
-		m_states_num = 0;
-		m_last_state = ControllerState();
+	if (!m_connected && new_connected) {
+		m_connected_count++;
 	}
+	m_active_id     = new_active_id;
+	m_connected     = new_connected;
+	m_state         = {};
+	m_states_num    = 0;
+	m_first_state   = 0;
+	m_next_touch_id = 1;
 }
 
-ControllerState GameController::GetLastState() const {
-	if (m_states_num == 0) {
-		return m_last_state;
-	}
-
-	auto last = (m_first_state + m_states_num - 1) % STATES_MAX;
-
-	return m_states[last];
-}
-
-void GameController::AddState(const ControllerState& state) {
+void GameController::AddState() {
 	if (m_states_num >= STATES_MAX) {
 		m_states_num  = STATES_MAX - 1;
 		m_first_state = (m_first_state + 1) % STATES_MAX;
 	}
 
-	auto index = (m_first_state + m_states_num) % STATES_MAX;
-
-	m_states[index] = state;
-	m_last_state    = state;
-
-	m_private[index].obtained = false;
-
+	const auto index  = (m_first_state + m_states_num) % STATES_MAX;
+	m_states[index]   = m_state;
+	m_obtained[index] = false;
 	m_states_num++;
 }
 
@@ -218,17 +206,11 @@ void GameController::Button(int id, uint32_t button, bool down) {
 
 	// The keyboard shares the player-1 pad with the active gamepad.
 	if (m_active_id == id || id == HOST_INPUT_CONTROLLER_ID) {
-		auto state = GetLastState();
+		m_state.time = LibKernel::KernelGetProcessTime();
 
-		state.time = LibKernel::KernelGetProcessTime();
+		m_state.buttons = down ? m_state.buttons | button : m_state.buttons & ~button;
 
-		if (down) {
-			state.buttons |= button;
-		} else {
-			state.buttons &= ~button;
-		}
-
-		AddState(state);
+		AddState();
 	}
 }
 
@@ -236,33 +218,25 @@ void GameController::Axis(int id, Controller::Axis axis, int value) {
 	Common::LockGuard lock(m_mutex);
 
 	if (m_active_id == id || id == HOST_INPUT_CONTROLLER_ID) {
-		auto state = GetLastState();
-
-		state.time = LibKernel::KernelGetProcessTime();
+		m_state.time = LibKernel::KernelGetProcessTime();
 
 		int axis_id = static_cast<int>(axis);
 
 		EXIT_IF(axis_id < 0 || axis_id >= static_cast<int>(Controller::Axis::AxisMax));
 
-		state.axes[axis_id] = value;
+		m_state.axes[axis_id] = value;
 
+		uint32_t trigger = 0;
 		if (axis == Controller::Axis::TriggerLeft) {
-			if (value > 0) {
-				state.buttons |= PAD_BUTTON_L2;
-			} else {
-				state.buttons &= ~PAD_BUTTON_L2;
-			}
+			trigger = PAD_BUTTON_L2;
+		} else if (axis == Controller::Axis::TriggerRight) {
+			trigger = PAD_BUTTON_R2;
+		}
+		if (trigger != 0) {
+			m_state.buttons = value > 0 ? m_state.buttons | trigger : m_state.buttons & ~trigger;
 		}
 
-		if (axis == Controller::Axis::TriggerRight) {
-			if (value > 0) {
-				state.buttons |= PAD_BUTTON_R2;
-			} else {
-				state.buttons &= ~PAD_BUTTON_R2;
-			}
-		}
-
-		AddState(state);
+		AddState();
 	}
 }
 
@@ -270,21 +244,45 @@ void GameController::RightStick(int id, int x, int y) {
 	Common::LockGuard lock(m_mutex);
 
 	if (m_active_id == id || id == HOST_INPUT_CONTROLLER_ID) {
-		auto state                                 = GetLastState();
-		state.time                                 = LibKernel::KernelGetProcessTime();
-		state.axes[static_cast<int>(Axis::RightX)] = x;
-		state.axes[static_cast<int>(Axis::RightY)] = y;
-		AddState(state);
+		m_state.time                                 = LibKernel::KernelGetProcessTime();
+		m_state.axes[static_cast<int>(Axis::RightX)] = x;
+		m_state.axes[static_cast<int>(Axis::RightY)] = y;
+		AddState();
+	}
+}
+
+void GameController::TouchPad(int id, int finger, bool down, float x, float y) {
+	if (finger < 0 || finger >= 2) {
+		return;
+	}
+
+	Common::LockGuard lock(m_mutex);
+	if (m_active_id == id || id == HOST_INPUT_CONTROLLER_ID) {
+		auto& touch  = m_state.touch[finger];
+		m_state.time = LibKernel::KernelGetProcessTime();
+		if (down && !touch.down) {
+			touch.id        = m_next_touch_id;
+			m_next_touch_id = m_next_touch_id == 127 ? 1 : m_next_touch_id + 1;
+		}
+		touch.down = down;
+		touch.x    = static_cast<uint16_t>(std::clamp(x, 0.0f, 1.0f) * 1920.0f);
+		touch.y    = static_cast<uint16_t>(std::clamp(y, 0.0f, 1.0f) * 943.0f);
+		if (id == HOST_INPUT_CONTROLLER_ID) {
+			m_state.buttons = down ? m_state.buttons | PAD_BUTTON_TOUCH_PAD
+			                       : m_state.buttons & ~PAD_BUTTON_TOUCH_PAD;
+		}
+		AddState();
 	}
 }
 
 void GameController::ResetInputState() {
 	Common::LockGuard lock(m_mutex);
-	ControllerState   state {};
-	state.time    = LibKernel::KernelGetProcessTime();
-	m_states_num  = 0;
-	m_first_state = 0;
-	AddState(state);
+	m_state         = {};
+	m_state.time    = LibKernel::KernelGetProcessTime();
+	m_states_num    = 0;
+	m_first_state   = 0;
+	m_next_touch_id = 1;
+	AddState();
 }
 
 void GameController::SetVibration(uint8_t large_motor, uint8_t small_motor) {
@@ -325,7 +323,7 @@ void GameController::ReadState(ControllerState* state, bool* flag, int* count) {
 
 	*flag  = m_connected;
 	*count = m_connected_count;
-	*state = GetLastState();
+	*state = m_state;
 }
 
 int GameController::ReadStates(ControllerState* states, int states_num, bool* flag, int* count) {
@@ -348,8 +346,8 @@ int GameController::ReadStates(ControllerState* states, int states_num, bool* fl
 					break;
 				}
 				auto index = (m_first_state + i) % STATES_MAX;
-				if (!m_private[index].obtained) {
-					m_private[index].obtained = true;
+				if (!m_obtained[index]) {
+					m_obtained[index] = true;
 
 					states[ret_num++] = m_states[index];
 				}
@@ -390,6 +388,12 @@ void ControllerRightStick(int id, int x, int y) {
 	g_controller->RightStick(id, x, y);
 }
 
+void ControllerTouchPad(int id, int finger, bool down, float x, float y) {
+	EXIT_IF(g_controller == nullptr);
+
+	g_controller->TouchPad(id, finger, down, x, y);
+}
+
 void ControllerResetInputState() {
 	EXIT_IF(g_controller == nullptr);
 	g_controller->ResetInputState();
@@ -508,7 +512,7 @@ int KYTY_SYSV_ABI PadGetControllerInformation(int handle, PadControllerInformati
 	info->stick_dead_zone_left  = controller_get_axis(-32768, 32767, 8000) - 128;
 	info->stick_dead_zone_right = controller_get_axis(-32768, 32767, 8000) - 128;
 	info->connection_type       = 0;
-	info->connected_count       = pad_connected_count_to_u8(connected_count);
+	info->connected_count       = static_cast<uint8_t>(std::min(connected_count, 255));
 	info->connected             = connected;
 	info->device_class          = 0;
 

src/libs/controller.h

@@ -51,7 +51,13 @@ struct PadLightBarParam;
 
 inline int controller_get_axis(int min, int max, int value) {
 	int v = (255 * (value - min)) / (max - min);
-	return (v < 0 ? 0 : (v > 255 ? 255 : v));
+	if (v < 0) {
+		return 0;
+	}
+	if (v > 255) {
+		return 255;
+	}
+	return v;
 }
 
 void ControllerConnect(int id);
@@ -59,6 +65,7 @@ void ControllerDisconnect(int id);
 void ControllerButton(int id, uint32_t button, bool down);
 void ControllerAxis(int id, Axis axis, int value);
 void ControllerRightStick(int id, int x, int y);
+void ControllerTouchPad(int id, int finger, bool down, float x, float y);
 void ControllerResetInputState();
 
 int KYTY_SYSV_ABI PadInit();

src/libs/libAgcDriver.cpp

@@ -37,7 +37,6 @@ LIB_DEFINE(InitAgcDriver_1) {
 	LIB_FUNC("V++UgBtQhn0", Gen5::AgcGetDataPacketPayloadAddress);
 	LIB_FUNC("s+VGAMDQ0AQ", Gen5::AgcGetDataPacketPayloadRange);
 	LIB_FUNC("fPSCdQxgpSw", Gen5::AgcWriteDataPatchSetAddressOrOffset);
-	LIB_FUNC("bxca0BK4FNg", Gen5::AgcUnknownJumpPatchSetTarget);
 	LIB_FUNC("2BS4EtAaF28", Gen5::AgcJumpPatchSetTarget);
 	LIB_FUNC("h9z6+0hEydk", Gen5::AgcSuspendPoint);
 	LIB_FUNC("qj7QZpgr9Uw", Gen5::AgcDcbContextStateOp);

src/libs/libAmpr.cpp

@@ -2665,6 +2665,7 @@ LIB_DEFINE(InitAmpr_1_Ampr) {
 	LIB_FUNC("dXPaz65HNmk", Ampr::CommandBufferPushMarker);
 	LIB_FUNC("mv0O8Zg0woU", Ampr::CommandBufferPopMarker);
 	LIB_FUNC("DLfoNxTFNVk", Ampr::CommandBufferWaitOnAddress);
+	LIB_FUNC("V7GQTEeUfhw", Ampr::CommandBufferWaitOnAddress);
 	LIB_FUNC("cQb8Zr8Q0Y0", Ampr::CommandBufferWaitOnCounter);
 	LIB_FUNC("j0+3uJMxYJY", Ampr::CommandBufferWriteAddress);
 	LIB_FUNC("sJXyWHjP-F8", Ampr::CommandBufferWriteAddressOnCompletion);

src/libs/libNet.cpp

@@ -635,6 +635,7 @@ LIB_DEFINE(InitNet_1_Http) {
 	LIB_FUNC("P6A3ytpsiYc", Http::HttpDeleteConnection);
 	LIB_FUNC("tsGVru3hCe8", Http::HttpCreateRequest);
 	LIB_FUNC("Aeu5wVKkF9w", Http::HttpCreateRequest);
+	LIB_FUNC("rGNm+FjIXKk", Http::HttpCreateRequest2);
 	LIB_FUNC("Cnp77podkCU", Http::HttpCreateRequestWithURL2);
 	LIB_FUNC("qe7oZ+v4PWA", Http::HttpDeleteRequest);
 	LIB_FUNC("PTiFIUxCpJc", Http::HttpSetRequestContentLength);

src/libs/network.cpp

@@ -1055,8 +1055,7 @@ static int ConvertGuestSockaddr(const void* addr, uint32_t addrlen, sockaddr_sto
 	return 0;
 }
 
-#if defined(_WIN32)
-static int ConvertHostSockaddr(const sockaddr_storage* addr, int addrlen, void* out,
+static int ConvertHostSockaddr(const sockaddr_storage* addr, SocketLength addrlen, void* out,
                                uint32_t* out_len) {
 	EXIT_IF(addr == nullptr);
 	EXIT_IF(out_len == nullptr);
@@ -1066,7 +1065,8 @@ static int ConvertHostSockaddr(const sockaddr_storage* addr, int addrlen, void*
 		return -1;
 	}
 
-	if (addr->ss_family != AF_INET || addrlen < static_cast<int>(sizeof(sockaddr_in))) {
+	if (addr->ss_family != AF_INET ||
+	    addrlen < static_cast<SocketLength>(sizeof(sockaddr_in))) {
 		*Posix::GetErrorAddr() = Posix::POSIX_EOPNOTSUPP;
 		return -1;
 	}
@@ -1088,7 +1088,6 @@ static int ConvertHostSockaddr(const sockaddr_storage* addr, int addrlen, void*
 	*out_len = guest_addrlen;
 	return 0;
 }
-#endif
 
 static bool GetSocketBackend(int guest_fd, NativeSocket* out) {
 	EXIT_IF(out == nullptr);
@@ -1851,19 +1850,13 @@ int KYTY_SYSV_ABI Getsockname(int s, void* addr, uint32_t* addrlen) {
 		return -1;
 	}
 
-#if defined(_WIN32)
 	sockaddr_storage host_addr {};
-	int              host_addrlen = sizeof(host_addr);
-	if (::getsockname(socket, reinterpret_cast<sockaddr*>(&host_addr), &host_addrlen) ==
-	    SOCKET_ERROR) {
+	SocketLength     host_addrlen = sizeof(host_addr);
+	if (::getsockname(socket, reinterpret_cast<sockaddr*>(&host_addr), &host_addrlen) != 0) {
 		return SetHostSocketError();
 	}
 
 	return ConvertHostSockaddr(&host_addr, host_addrlen, addr, addrlen);
-#else
-	*Posix::GetErrorAddr() = Posix::POSIX_ENOSYS;
-	return -1;
-#endif
 }
 
 int KYTY_SYSV_ABI Getsockopt(int s, int level, int optname, void* optval, uint32_t* optlen) {
@@ -2818,6 +2811,27 @@ int KYTY_SYSV_ABI HttpCreateRequest(int conn_id, int method, const char* path,
 	return id.ToInt();
 }
 
+int KYTY_SYSV_ABI HttpCreateRequest2(int conn_id, const char* method, const char* path,
+                                     uint64_t content_length) {
+	PRINT_NAME();
+
+	LOGF("\t conn_id        = %d\n"
+	     "\t method         = %s\n"
+	     "\t path           = %s\n"
+	     "\t content_length = %" PRIu64 "\n",
+	     conn_id, method, path, content_length);
+
+	EXIT_IF(g_net == nullptr);
+
+	if (method == nullptr || path == nullptr) {
+		return HTTP_ERROR_INVALID_VALUE;
+	}
+
+	auto id =
+	    g_net->HttpCreateRequestWithURL2(Network::Id(conn_id), method, path, content_length);
+	return id.IsValid() ? id.ToInt() : HTTP_ERROR_OUT_OF_MEMORY;
+}
+
 int KYTY_SYSV_ABI HttpCreateRequestWithURL2(int conn_id, const char* method, const char* url,
                                             uint64_t content_length) {
 	PRINT_NAME();

src/libs/network.h

@@ -106,6 +106,8 @@ int KYTY_SYSV_ABI HttpCreateConnectionWithURL(int tmpl_id, const char* url, int
 int KYTY_SYSV_ABI HttpDeleteConnection(int conn_id);
 int KYTY_SYSV_ABI HttpCreateRequest(int conn_id, int method, const char* path,
                                     uint64_t content_length);
+int KYTY_SYSV_ABI HttpCreateRequest2(int conn_id, const char* method, const char* path,
+                                     uint64_t content_length);
 int KYTY_SYSV_ABI HttpCreateRequestWithURL2(int conn_id, const char* method, const char* url,
                                             uint64_t content_length);
 int KYTY_SYSV_ABI HttpSetRequestContentLength(int request_id, uint64_t content_length);

src/libs/padData.h

@@ -5,46 +5,50 @@
 
 namespace Libs::Controller {
 
+struct PadTouch {
+	uint16_t x;
+	uint16_t y;
+	uint8_t  id;
+	uint8_t  reserve[3];
+};
+
+struct PadTouchData {
+	uint8_t  touch_num;
+	uint8_t  reserve[3];
+	uint32_t reserve1;
+	PadTouch touch[2];
+};
+
 struct PadData {
-	uint32_t buttons;
-	uint8_t  left_stick_x;
-	uint8_t  left_stick_y;
-	uint8_t  right_stick_x;
-	uint8_t  right_stick_y;
-	uint8_t  analog_buttons_l2;
-	uint8_t  analog_buttons_r2;
-	uint8_t  padding[2];
-	float    orientation_x;
-	float    orientation_y;
-	float    orientation_z;
-	float    orientation_w;
-	float    acceleration_x;
-	float    acceleration_y;
-	float    acceleration_z;
-	float    angular_velocity_x;
-	float    angular_velocity_y;
-	float    angular_velocity_z;
-	uint8_t  touch_data_touch_num;
-	uint8_t  touch_data_reserve[3];
-	uint32_t touch_data_reserve1;
-	uint16_t touch_data_touch0_x;
-	uint16_t touch_data_touch0_y;
-	uint8_t  touch_data_touch0_id;
-	uint8_t  touch_data_touch0_reserve[3];
-	uint16_t touch_data_touch1_x;
-	uint16_t touch_data_touch1_y;
-	uint8_t  touch_data_touch1_id;
-	uint8_t  touch_data_touch1_reserve[3];
-	bool     connected;
-	uint64_t timestamp;
-	uint32_t extension_unit_data_extension_unit_id;
-	uint8_t  extension_unit_data_reserve[1];
-	uint8_t  extension_unit_data_data_length;
-	uint8_t  extension_unit_data_data[10];
-	uint8_t  connected_count;
-	uint8_t  reserve[2];
-	uint8_t  device_unique_data_len;
-	uint8_t  device_unique_data[12];
+	uint32_t     buttons;
+	uint8_t      left_stick_x;
+	uint8_t      left_stick_y;
+	uint8_t      right_stick_x;
+	uint8_t      right_stick_y;
+	uint8_t      analog_buttons_l2;
+	uint8_t      analog_buttons_r2;
+	uint8_t      padding[2];
+	float        orientation_x;
+	float        orientation_y;
+	float        orientation_z;
+	float        orientation_w;
+	float        acceleration_x;
+	float        acceleration_y;
+	float        acceleration_z;
+	float        angular_velocity_x;
+	float        angular_velocity_y;
+	float        angular_velocity_z;
+	PadTouchData touch_data;
+	bool         connected;
+	uint64_t     timestamp;
+	uint32_t     extension_unit_data_extension_unit_id;
+	uint8_t      extension_unit_data_reserve[1];
+	uint8_t      extension_unit_data_data_length;
+	uint8_t      extension_unit_data_data[10];
+	uint8_t      connected_count;
+	uint8_t      reserve[2];
+	uint8_t      device_unique_data_len;
+	uint8_t      device_unique_data[12];
 };
 
 static_assert(sizeof(PadData) == 120);

src/loader/runtimeLinker.cpp

@@ -1166,7 +1166,6 @@ static void PatchProgram(Program* program, uint64_t address, uint64_t size) {
 				prefix_count++;
 			}
 
-			const size_t inst_size = prefix_count + Jit::Call9::GetSize();
 			if (inst_ptr + Jit::Call9::GetSize() > start_ptr + size) {
 				break;
 			}
@@ -1175,20 +1174,18 @@ static void PatchProgram(Program* program, uint64_t address, uint64_t size) {
 			if (memcmp(inst_ptr, tls_pattern, 3) == 0 && (modrm & 0xc7u) == 0x04u &&
 			    inst_ptr[4] == tls_pattern[4] &&
 			    *reinterpret_cast<const uint32_t*>(inst_ptr + 5) == 0) {
-				LOGF("Patch tls at addr: [%016" PRIx64 "]\n", reinterpret_cast<uint64_t>(ptr));
+				LOGF("Patch tls at addr: [%016" PRIx64 "]\n",
+				     reinterpret_cast<uint64_t>(inst_ptr));
 
 				const auto reg = (modrm >> 3u) & 7u;
 				EXIT_NOT_IMPLEMENTED(reg == 4u);
 
-				auto* code = new (ptr) Jit::Call9;
+				// A raw scan can encounter a 0x66 in the preceding instruction.
+				auto* code = new (inst_ptr) Jit::Call9;
 				code->SetFunc(reg == 0
 				                  ? program->tls.handler_vaddr
 				                  : program->tls.handler_vaddr + Jit::TlsRegStub::GetOffset(reg));
-				if (inst_size > Jit::Call9::GetSize()) {
-					std::memset(ptr + Jit::Call9::GetSize(), 0x90,
-					            inst_size - Jit::Call9::GetSize());
-				}
-				ptr += inst_size - 1;
+				ptr += prefix_count + Jit::Call9::GetSize() - 1;
 			}
 		}
 	}

tests/MemoryTrackerTests.cpp

@@ -29,9 +29,11 @@
 
 namespace {
 
+using Libs::Graphics::GuestRange;
 using Libs::Graphics::MemoryTracker;
 using Libs::Graphics::PageManager;
 using Libs::Graphics::RangeSet;
+using Libs::Graphics::TRACKER_ADDRESS_SIZE;
 
 void Check(bool value, const char *text) {
   if (!value) {
@@ -208,6 +210,27 @@ void TestRangeSet() {
         "range set subtraction did not preserve both exact tails");
 }
 
+void TestGuestRange() {
+  constexpr GuestRange empty{};
+  constexpr GuestRange first_byte{1, 1};
+  constexpr GuestRange last_byte{TRACKER_ADDRESS_SIZE - 1, 1};
+
+  static_assert(empty.Empty() && !empty.Valid() && empty.ValidOrEmpty());
+  static_assert(!first_byte.Empty() && first_byte.Valid() &&
+                first_byte.ValidOrEmpty() && first_byte.End() == 2);
+  static_assert(last_byte.Valid() && last_byte.End() == TRACKER_ADDRESS_SIZE);
+
+  Check(!GuestRange{0, 1}.Empty() && !GuestRange{0, 1}.ValidOrEmpty(),
+        "zero-address nonempty guest range is rejected");
+  Check(!GuestRange{1, 0}.Empty() && !GuestRange{1, 0}.ValidOrEmpty(),
+        "nonzero-address empty guest range is rejected");
+  Check(!GuestRange{TRACKER_ADDRESS_SIZE, 1}.Valid(),
+        "first address beyond the guest range is rejected");
+  Check(!GuestRange{TRACKER_ADDRESS_SIZE - 1, 2}.Valid(),
+        "guest range crossing the address-space end is rejected");
+  Check(!GuestRange{UINT64_MAX, 2}.Valid(), "wrapping guest range is rejected");
+}
+
 void TestQueriesDoNotRequireMappedOwnership() {
   constexpr uint64_t address = 0x0000000203000000ull;
   TrackerHarness harness;
@@ -829,6 +852,7 @@ int main(int argc, char **argv) {
   if (argc == 3 && std::strcmp(argv[1], "--death") == 0) {
     RunDeathCase(argv[2]);
   }
+  TestGuestRange();
   TestRangeSet();
   TestQueriesDoNotRequireMappedOwnership();
   TestConcurrentRegionPublication();

tests/ShaderRecompilerComputeTests.cpp

@@ -357,10 +357,23 @@ struct TextureCacheTestAccess {
     return cache.TryDownloadImage(id);
   }
 
+  static void RegisterHtileMeta(TextureCache &cache, uint64_t address) {
+    std::lock_guard lock(cache.m_lock);
+    auto &metadata = cache.m_surface_metas[address];
+    metadata.type = TextureCache::MetaDataInfo::Type::HTile;
+    metadata.clear_mask = 0;
+  }
+
   static TileManager &Tiler(TextureCache &cache) { return cache.m_tiler; }
 };
 
 struct RenderExecutorTestAccess {
+  static bool TryConsumeComputeMetaClear(RenderExecutor &executor,
+                                         const ShaderComputeInputInfo &input,
+                                         const CommandBuffer &buffer) {
+    return executor.TryConsumeComputeMetaClear(input, buffer);
+  }
+
   static TextureBinding
   ResolveTexture(RenderExecutor &executor,
                  const ShaderRecompiler::IR::ImageResource &resource,
@@ -1209,7 +1222,7 @@ CompiledShader CompileCase(const TestCase &test) {
   options.stage = ShaderType::Compute;
   options.dump_ir = true;
   options.input_info.compute = &test.compute_info;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
   options.read_memory = ReadTestMemory;
   options.read_memory_data = const_cast<std::vector<u32> *>(&test.initial);
   options.scratch_dwords = test.compute_info.scratch_size_dwords;
@@ -1349,7 +1362,7 @@ CompiledShader CompileFragmentCase(const GraphicsCase &test) {
   options.stage = ShaderType::Pixel;
   options.dump_ir = false;
   options.input_info.pixel = &pixel_info;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(test.fragment_code, options);
   Require(test.name, "SPIR-V emit", !result.spirv.empty(),
@@ -3778,6 +3791,92 @@ public:
     std::printf("[host]    %-32s ok\n", name);
   }
 
+  void CheckComputeMetaClearClassification() {
+    constexpr const char *name = "ComputeMetaClearClassification";
+    constexpr uint64_t read_only_meta = 0x0000000204201f00ull;
+    constexpr uint64_t read_write_meta = 0x0000000204202000ull;
+    constexpr uint64_t write_only_meta = 0x0000000204202100ull;
+    EnsureRuntimeContext();
+    RenderContext context(m_runtime_context);
+    auto &scheduler = context.GetCommandScheduler();
+    HW::Context registers{};
+    HW::UserConfig user_config{};
+    HW::Shader shaders{};
+    scheduler.Begin(registers, user_config, shaders);
+    auto &texture_cache = context.GetTextureCache();
+    TextureCacheTestAccess::RegisterHtileMeta(texture_cache, read_only_meta);
+    TextureCacheTestAccess::RegisterHtileMeta(texture_cache, read_write_meta);
+    TextureCacheTestAccess::RegisterHtileMeta(texture_cache, write_only_meta);
+    Require(name, "HTile fixture",
+            texture_cache.IsMeta(read_only_meta) &&
+                texture_cache.IsMeta(read_write_meta) &&
+                texture_cache.IsMeta(write_only_meta) &&
+                !texture_cache.IsMetaCleared(read_only_meta, 0) &&
+                !texture_cache.IsMetaCleared(read_write_meta, 0) &&
+                !texture_cache.IsMetaCleared(write_only_meta, 0),
+            "test HTile entries were not registered in their initial state");
+
+    const auto MakeInput = [](uint64_t address, bool read, bool written) {
+      auto program = std::make_shared<ShaderRecompiler::IR::Program>();
+      program->stage = ShaderType::Compute;
+      ShaderRecompiler::IR::BufferResource resource{};
+      resource.read = read;
+      resource.written = written;
+      resource.formatted = true;
+      program->info.buffers.push_back(resource);
+
+      ShaderBufferResource descriptor{};
+      descriptor.UpdateAddress48(address);
+      descriptor.fields[1] |= 16u << 16u;
+      descriptor.fields[2] = 8;
+      descriptor.fields[3] =
+          DstSel(4, 5, 6, 7) |
+          (static_cast<uint32_t>(Prospero::BufferFormat::k32_32_32_32UInt)
+           << 12u);
+      ShaderRecompiler::IR::DescriptorValue value{};
+      value.dword_count = 4;
+      std::copy_n(descriptor.fields, value.dword_count, value.dwords.begin());
+      auto snapshot =
+          std::make_shared<ShaderRecompiler::IR::ResourceSnapshot>();
+      snapshot->buffers.push_back(value);
+
+      ShaderComputeInputInfo input{};
+      input.stage.program = std::move(program);
+      input.stage.resources = std::move(snapshot);
+      return input;
+    };
+    auto &executor = context.GetRenderExecutor();
+    auto &command = scheduler.Current();
+    const auto read_only_input = MakeInput(read_only_meta, true, false);
+    const bool read_only_consumed =
+        RenderExecutorTestAccess::TryConsumeComputeMetaClear(
+            executor, read_only_input, command);
+    Require(name, "read-only dispatch preserved",
+            !read_only_consumed &&
+                !texture_cache.IsMetaCleared(read_only_meta, 0),
+            "a metadata read-only dispatch changed logical clear state");
+
+    const auto read_write_input = MakeInput(read_write_meta, true, true);
+    const bool read_write_consumed =
+        RenderExecutorTestAccess::TryConsumeComputeMetaClear(
+            executor, read_write_input, command);
+    Require(name, "read/write dispatch preserved",
+            !read_write_consumed &&
+                !texture_cache.IsMetaCleared(read_write_meta, 0),
+            "a metadata read/modify/write dispatch was replaced by a clear");
+
+    const auto write_only_input = MakeInput(write_only_meta, false, true);
+    const bool write_only_consumed =
+        RenderExecutorTestAccess::TryConsumeComputeMetaClear(
+            executor, write_only_input, command);
+    Require(name, "write-only fill consumed",
+            write_only_consumed &&
+                texture_cache.IsMetaCleared(write_only_meta, 0),
+            "a metadata write-only fill was not consumed as a clear");
+    scheduler.Finish();
+    std::printf("[host]    %-32s ok\n", name);
+  }
+
   void CheckUnifiedTextureCacheFlow() {
     constexpr const char *name = "UnifiedTextureCacheFlow";
     constexpr uintptr_t base = 0x0000000200600000ull;
@@ -7060,8 +7159,7 @@ public:
       registers.SetColorAttrib3(0,
                                 {.tile_mode = Prospero::TileMode::kRenderTarget,
                                  .dimension = 0,
-                                 .cmask_pipe_aligned = true,
-                                 .dcc_pipe_aligned = true});
+                                 .metadata_pipe_aligned = true});
       registers.SetRenderTargetMask(0x0f);
       scheduler.Begin(registers, user_config, shaders);
 
@@ -7094,7 +7192,7 @@ public:
                   rendering.width == width && rendering.height == height &&
                   rendering.num_layers == 1 &&
                   rendering.num_color_attachments == 1,
-              "SDK dimension=0 did not create a 512x1 Vulkan attachment");
+              "dimension=0 did not create a 512x1 Vulkan attachment");
 
       RenderExecutorTestAccess::ResetBindings(executor);
       resources.UnmapMemory(base, allocation_size);
@@ -7153,8 +7251,7 @@ public:
                                 {.depth = 31,
                                  .tile_mode = Prospero::TileMode::kRenderTarget,
                                  .dimension = 2,
-                                 .cmask_pipe_aligned = true,
-                                 .dcc_pipe_aligned = true});
+                                 .metadata_pipe_aligned = true});
       constexpr uint64_t dcc_address = base + allocation_size - 0x1000;
       registers.SetColorDccAddr(0, {.addr = dcc_address});
       registers.SetColorClearWord0(0, {.word0 = 0});
@@ -7198,7 +7295,7 @@ public:
               static_cast<bool>(image.backing.flags &
                                 vk::ImageCreateFlagBits::e2DArrayCompatible) &&
               image.usage.render_target && image.IsGpuModified(),
-          "dimension=2/depth=31 did not create the SDK-defined 32x32x32 "
+          "dimension=2/depth=31 did not create the expected 32x32x32 "
           "backing and 2D "
           "attachment slice");
 
@@ -7367,8 +7464,7 @@ public:
       registers.SetColorAttrib2(0, {.height = 255, .width = 255});
       registers.SetColorAttrib3(0, {.tile_mode = tile_case.tile,
                                     .dimension = 1,
-                                    .cmask_pipe_aligned = true,
-                                    .dcc_pipe_aligned = true});
+                                    .metadata_pipe_aligned = true});
       registers.SetRenderTargetMask(0x0f);
       scheduler.Begin(registers, user_config, shaders);
 
@@ -7477,8 +7573,7 @@ public:
       registers.SetColorAttrib2(0, {.height = 63, .width = 63});
       registers.SetColorAttrib3(0, {.tile_mode = Prospero::TileMode::kDepth,
                                     .dimension = 1,
-                                    .cmask_pipe_aligned = true,
-                                    .dcc_pipe_aligned = true});
+                                    .metadata_pipe_aligned = true});
       registers.SetRenderTargetMask(0x0f);
       scheduler.Begin(registers, user_config, shaders);
 
@@ -8362,12 +8457,17 @@ public:
       RenderExecutorTestAccess::ResetBindings(executor);
 
       constexpr uint64_t depth_only_address = base + 0x140000;
+      constexpr uint64_t depth_only_htile_address = base + 0x150000;
+      constexpr uint32_t captured_depth_only_z_info = 0x22900983u;
+      constexpr uint32_t captured_depth_only_stencil_info = 0x00100980u;
       HW::DepthRenderTarget depth_only_target{};
-      depth_only_target.z_info.format = Prospero::DepthFormat::kZ32F;
-      depth_only_target.z_info.z_compare_base = Prospero::ZCompareBase::kZMax;
-      depth_only_target.stencil_info.htile_stencil_disabled = true;
+      depth_only_target.z_info =
+          HW::DepthZInfo::Decode(captured_depth_only_z_info);
+      depth_only_target.stencil_info =
+          HW::DepthStencilInfo::Decode(captured_depth_only_stencil_info);
       depth_only_target.z_read_base_addr = depth_only_address;
       depth_only_target.z_write_base_addr = depth_only_address;
+      depth_only_target.htile_data_base_addr = depth_only_htile_address;
       depth_only_target.size = {63, 63, true};
       registers.SetDepthRenderTarget(depth_only_target);
       HW::DepthControl depth_only_control{};
@@ -8386,6 +8486,7 @@ public:
       Require(
           name, "depth-only target with stale stencil state",
           depth_only.image_id && depth_only.format == vk::Format::eD32Sfloat &&
+              depth_only.htile &&
               depth_only.depth_test_enable && depth_only.depth_write_enable &&
               depth_only.depth_clear_enable &&
               !depth_only.stencil_test_enable &&
@@ -8393,14 +8494,16 @@ public:
               depth_only.stencil_buffer_vaddr == 0 &&
               depth_only.stencil_buffer_size == 0 &&
               depth_only.desc.info.stencil.Empty() &&
+              depth_only.htile_buffer_vaddr == depth_only_htile_address &&
+              depth_only.desc.info.metadata.range.address ==
+                  depth_only_htile_address &&
               depth_only.depth_buffer_size != 0 &&
               depth_only_address + depth_only.depth_buffer_size <=
                   base + allocation_size &&
               depth_only.vaddr_num == 1 &&
               depth_only.AttachmentWriteAspects() ==
                   vk::ImageAspectFlagBits::eDepth,
-          "raw stencil test or clear state leaked into a depth-only "
-          "attachment");
+          "inactive stencil state changed a depth-only HTile attachment");
       RenderExecutorTestAccess::ResetBindings(executor);
 
       const auto sampled_width = depth_only.desc.info.extent.width - 1u;
@@ -12737,160 +12840,6 @@ TestCase ScalarNotB64UpdatesScc() {
            O::S_ENDPGM}};
 }
 
-TestCase ScalarMovB64PreservesRawSpecialData() {
-  using O = ShaderOpcode;
-
-  std::vector<u32> code;
-  AppendSMovLiteral(&code, 0, 0x09500000u);
-  AppendSMovLiteral(&code, 1, 0x81234567u);
-  code.push_back(EncodeSop1(0x04, 106, 0)); // s_mov_b64 vcc, s[0:1]
-  code.push_back(EncodeSop1(0x04, 2, 106)); // s_mov_b64 s[2:3], vcc
-  code.push_back(EncodeSop1(0x08, 4, 2));   // s_not_b64 s[4:5], s[2:3]
-  code.push_back(EncodeSop2(0x0a, 6, InlineU32(1), InlineU32(0)));
-
-  AppendStoreSgprPair(&code, 2, 0);
-  AppendStoreSgprPair(&code, 4, 2);
-  AppendStoreSgpr(&code, 6, 4);
-
-  AppendSMovLiteral(&code, 8, 0x09500001u);
-  AppendSMovLiteral(&code, 9, 0x81234567u);
-  code.push_back(EncodeSop1(0x04, 126, 8));  // s_mov_b64 exec, s[8:9]
-  code.push_back(EncodeSop1(0x04, 10, 126)); // s_mov_b64 s[10:11], exec
-  code.push_back(EncodeSop1(0x08, 12, 10));  // s_not_b64 s[12:13], s[10:11]
-  code.push_back(EncodeSop2(0x0a, 14, InlineU32(1), InlineU32(0)));
-
-  AppendStoreSgprPair(&code, 10, 5);
-  AppendStoreSgprPair(&code, 12, 7);
-  AppendStoreSgpr(&code, 14, 9);
-  AppendEnd(&code);
-
-  return {"ScalarMovB64PreservesRawSpecialData",
-          code,
-          {},
-          {0x09500000u, 0x81234567u, 0xf6afffffu, 0x7edcba98u, 1u,
-           0x09500001u, 0x81234567u, 0xf6affffeu, 0x7edcba98u, 1u},
-          {O::S_MOV_B32, O::S_MOV_B64, O::S_NOT_B64, O::S_CSELECT_B32,
-           O::V_MOV_B32, O::BUFFER_STORE_DWORD, O::S_ENDPGM}};
-}
-
-TestCase Scalar64RawSpecialOps() {
-  using O = ShaderOpcode;
-
-  std::vector<u32> code;
-  const auto capture_scc = [&](u32 dst) {
-    code.push_back(EncodeSop2(0x0a, dst, InlineU32(1), InlineU32(0)));
-  };
-
-  AppendSMovLiteral(&code, 0, 0x09500001u);
-  AppendSMovLiteral(&code, 1, 0x81234567u);
-  AppendSMovLiteral(&code, 2, 0xff00ff01u);
-  AppendSMovLiteral(&code, 3, 0x0f0f0f0fu);
-  code.push_back(EncodeSop1(0x04, 126, 0));
-  code.push_back(EncodeSop2(0x0f, 106, 126, 2));
-  code.push_back(EncodeSop1(0x04, 4, 106));
-  capture_scc(6);
-  code.push_back(EncodeSop1(0x08, 106, 106));
-  code.push_back(EncodeSop1(0x04, 8, 106));
-  capture_scc(10);
-
-  AppendSMovLiteral(&code, 12, 0x80000011u);
-  AppendSMovLiteral(&code, 13, 0x80000001u);
-  code.push_back(EncodeSop1(0x04, 106, 12));
-  code.push_back(EncodeSop1(0x0a, 106, 106));
-  code.push_back(EncodeSop1(0x04, 14, 106));
-  capture_scc(16);
-
-  code.push_back(EncodeSop1(0x0a, 26, 193u));
-  capture_scc(28);
-
-  code.push_back(EncodeSMovB32(18, InlineU32(1)));
-  code.push_back(EncodeSMovB32(19, InlineU32(0)));
-  code.push_back(EncodeSopc(0x06, InlineU32(1), InlineU32(1)));
-  code.push_back(EncodeSop2(0x0b, 126, 0, 18));
-  code.push_back(EncodeSop1(0x04, 20, 126));
-  code.push_back(EncodeSop1(0x08, 106, 126));
-  code.push_back(EncodeSop1(0x04, 22, 106));
-  capture_scc(24);
-  code.push_back(EncodeSop1(0x04, 126, 18));
-
-  AppendStoreSgprPair(&code, 4, 0);
-  AppendStoreSgpr(&code, 6, 2);
-  AppendStoreSgprPair(&code, 8, 3);
-  AppendStoreSgpr(&code, 10, 5);
-  AppendStoreSgprPair(&code, 14, 6);
-  AppendStoreSgpr(&code, 16, 8);
-  AppendStoreSgprPair(&code, 26, 9);
-  AppendStoreSgpr(&code, 28, 11);
-  AppendStoreSgprPair(&code, 20, 12);
-  AppendStoreSgprPair(&code, 22, 14);
-  AppendStoreSgpr(&code, 24, 16);
-  AppendEnd(&code);
-
-  return {"Scalar64RawSpecialOps",
-          code,
-          {},
-          {0x09000001u, 0x01030507u, 1, 0xf6fffffeu, 0xfefcfaf8u, 1,
-           0xf00000ffu, 0xf000000fu, 1, 0xffffffffu, 0xffffffffu, 1,
-           0x09500001u, 0x81234567u, 0xf6affffeu, 0x7edcba98u, 1},
-          {O::S_MOV_B32, O::S_MOV_B64, O::S_AND_B64, O::S_NOT_B64,
-           O::S_CSELECT_B32, O::S_WQM_B64, O::S_CMP_EQ_U32, O::S_CSELECT_B64,
-           O::V_MOV_B32, O::BUFFER_STORE_DWORD, O::S_ENDPGM}};
-}
-
-TestCase ScalarMovB32SpecialAliases() {
-  using O = ShaderOpcode;
-
-  std::vector<u32> code;
-  const auto capture_scc = [&](u32 dst) {
-    code.push_back(EncodeSop2(0x0a, dst, InlineU32(1), InlineU32(0)));
-  };
-
-  AppendSMovLiteral(&code, 0, 0x09500001u);
-  AppendSMovLiteral(&code, 1, 0x81234567u);
-  AppendSMovLiteral(&code, 2, 0xaaaaaaaau);
-  AppendSMovLiteral(&code, 3, 0xbbbbbbbbu);
-  code.push_back(EncodeSMovB32(4, InlineU32(1)));
-  code.push_back(EncodeSMovB32(5, InlineU32(0)));
-  code.push_back(EncodeSop1(0x04, 126, 0));
-  code.push_back(EncodeSop1(0x04, 106, 2));
-  code.push_back(EncodeSMovB32(107, 126));
-  code.push_back(EncodeSop1(0x04, 6, 106));
-  code.push_back(EncodeSop1(0x08, 8, 106));
-  capture_scc(10);
-
-  AppendSMovLiteral(&code, 12, 0x13579bdfu);
-  AppendSMovLiteral(&code, 13, 0x2468ace0u);
-  code.push_back(EncodeSop1(0x04, 106, 12));
-  code.push_back(EncodeSMovB32(126, 106));
-  code.push_back(EncodeSop1(0x04, 14, 126));
-  code.push_back(EncodeSop1(0x04, 126, 4));
-
-  code.push_back(EncodeVopc(0xc2, InlineU32(0), 0));
-  code.push_back(EncodeSMovB32(106, 106));
-  code.push_back(EncodeSop1(0x08, 16, 106));
-  capture_scc(18);
-  code.push_back(EncodeSop1(0x04, 20, 106));
-  code.push_back(EncodeSMovB32(20, 20));
-  code.push_back(EncodeSop1(0x08, 22, 20));
-  capture_scc(24);
-
-  AppendStoreSgprPair(&code, 6, 0);
-  AppendStoreSgprPair(&code, 8, 2);
-  AppendStoreSgpr(&code, 10, 4);
-  AppendStoreSgprPair(&code, 14, 5);
-  AppendStoreSgpr(&code, 18, 7);
-  AppendStoreSgpr(&code, 24, 8);
-  AppendEnd(&code);
-
-  return {"ScalarMovB32SpecialAliases",
-          code,
-          {},
-          {0xaaaaaaaau, 0x09500001u, 0x55555555u, 0xf6affffeu, 1, 0x13579bdfu,
-           0x81234567u, 0, 0},
-          {O::S_MOV_B32, O::S_MOV_B64, O::S_NOT_B64, O::S_CSELECT_B32,
-           O::V_CMP_EQ_U32, O::V_MOV_B32, O::BUFFER_STORE_DWORD, O::S_ENDPGM}};
-}
-
 TestCase ScalarQuadmaskB64() {
   using O = ShaderOpcode;
 
@@ -12956,6 +12905,36 @@ TestCase ScalarFlbitI32B64Gpu() {
            O::BUFFER_STORE_DWORD, O::S_ENDPGM}};
 }
 
+TestCase ScalarFlbitI32B32Gpu() {
+  using O = ShaderOpcode;
+
+  std::vector<u32> code;
+  AppendSMovLiteral(&code, 0, 0x00000000u);
+  code.push_back(EncodeSop1(0x15, 20, 0));
+
+  AppendSMovLiteral(&code, 0, 0x0000ccccu);
+  code.push_back(EncodeSop1(0x15, 21, 0));
+
+  AppendSMovLiteral(&code, 0, 0x80000000u);
+  code.push_back(EncodeSop1(0x15, 22, 0));
+
+  AppendSMovLiteral(&code, 0, 0x7fffffffu);
+  code.push_back(EncodeSop1(0x15, 23, 0));
+
+  AppendStoreSgpr(&code, 20, 0);
+  AppendStoreSgpr(&code, 21, 1);
+  AppendStoreSgpr(&code, 22, 2);
+  AppendStoreSgpr(&code, 23, 3);
+  AppendEnd(&code);
+
+  return {"ScalarFlbitI32B32Gpu",
+          code,
+          {},
+          {0xffffffffu, 16, 0, 1},
+          {O::S_MOV_B32, O::S_FLBIT_I32_B32, O::V_MOV_B32,
+           O::BUFFER_STORE_DWORD, O::S_ENDPGM}};
+}
+
 TestCase ScalarSaveExecOps() {
   using O = ShaderOpcode;
 
@@ -12989,7 +12968,8 @@ TestCase ScalarSaveExecOps() {
   return {"ScalarSaveExecOps",
           code,
           {},
-          {1, 0, 1, 0, 0xffffffffu, 0xffffffffu, 1, 1, 0, 1, 0, 0, 0},
+          // EXEC and saved EXEC values are invocation-local Booleans.
+          {1, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 0},
           {O::S_MOV_B32, O::S_AND_SAVEEXEC_B64, O::S_ORN2_SAVEEXEC_B64,
            O::S_ANDN1_SAVEEXEC_B64, O::S_AND_SAVEEXEC_B32,
            O::S_ANDN1_SAVEEXEC_B32, O::S_MOV_B64, O::V_MOV_B32,
@@ -13016,7 +12996,7 @@ TestCase ScalarOrn2SaveexecUsesSourceOrNotExec() {
   return {"ScalarOrn2SaveexecUsesSourceOrNotExec",
           code,
           {},
-          {0x0000000cu, 0x80000000u, 0xfffffff3u, 0x7fffffffu, 1},
+          {1, 0, 1, 0, 1},
           {O::S_MOV_B32, O::S_ORN2_SAVEEXEC_B64, O::S_MOV_B64, O::V_MOV_B32,
            O::BUFFER_STORE_DWORD, O::S_ENDPGM}};
 }
@@ -13677,6 +13657,32 @@ TestCase Vop2SdwaAshrrevCapturedWord0SignExtends() {
   return test;
 }
 
+TestCase Vop2SdwaLshrrevCapturedByte1Source() {
+  using O = ShaderOpcode;
+
+  std::vector<u32> code;
+  AppendSMovLiteral(&code, 106, 4);
+  AppendBufferLoadDword(&code, 8, 30);
+  code.push_back(0x2c1210f9u);
+  code.push_back(0x0186066au); // v_lshrrev_b32 v9, vcc_lo, v8.byte1
+  AppendStoreVgpr(&code, 9, 1);
+  AppendEnd(&code);
+
+  TestCase test;
+  test.name = "Vop2SdwaLshrrevCapturedByte1Source";
+  test.code = std::move(code);
+  test.initial = {0x12348678u, 0};
+  test.expected = {0x12348678u, 8};
+  test.opcodes = {O::S_MOV_B32, O::BUFFER_LOAD_DWORD, O::V_LSHRREV_B32,
+                  O::V_MOV_B32, O::BUFFER_STORE_DWORD, O::S_ENDPGM};
+  test.decoded_counts = {
+      {"V_LSHRREV_B32 v9, vcc_lo, v8.sdwa(sel=1,sext=0)", 1}};
+  test.ir_counts = {{" = BitFieldUExtract ", 1},
+                    {" = ShiftRightLogical32 ", 1}};
+  test.required_spirv = {"OpBitFieldUExtract", "OpShiftRightLogical"};
+  return test;
+}
+
 TestCase Vop2SdwaSubNcPreservesByteAndWordDestinations() {
   using O = ShaderOpcode;
 
@@ -15115,6 +15121,31 @@ TestCase VectorDppQuadPermuteReverse() {
   return test;
 }
 
+TestCase VectorDppRowXmask() {
+  using O = ShaderOpcode;
+
+  std::vector<u32> code;
+  AppendVMovU32(&code, 1, 100);
+  code.push_back(EncodeVop2(0x25, 2, 250, 1));
+  code.push_back(EncodeVop2Dpp(0, 0x164));
+  code.push_back(EncodeVop2(0x1a, 3, InlineU32(2), 0));
+  AppendBufferStoreDword(&code, 2, 3);
+  AppendEnd(&code);
+
+  TestCase test;
+  test.name = "VectorDppRowXmask";
+  test.code = code;
+  test.expected = {104, 105, 106, 107, 100, 101, 102, 103};
+  test.opcodes = {O::V_MOV_B32, O::V_ADD_NC_U32, O::V_LSHLREV_B32,
+                  O::BUFFER_STORE_DWORD, O::S_ENDPGM};
+  test.compute_info.threads_num[0] = 8;
+  test.compute_info.threads_num[1] = 1;
+  test.compute_info.threads_num[2] = 1;
+  test.compute_info.thread_ids_num = 1;
+  test.has_compute_info = true;
+  return test;
+}
+
 TestCase VectorDppBankMaskPreservesDestination() {
   using O = ShaderOpcode;
 
@@ -16061,6 +16092,36 @@ TestCase VectorVop3CompareGtU64OnGpu() {
   return test;
 }
 
+TestCase VectorVopcCompareLtU64OnGpu() {
+  using O = ShaderOpcode;
+
+  std::vector<u32> code;
+  AppendVMovU32(&code, 0, 5);
+  AppendVMovU32(&code, 1, 0);
+  AppendVMovU32(&code, 2, 1);
+
+  AppendSMovLiteral(&code, 4, 4);
+  AppendSMovLiteral(&code, 5, 0);
+  code.push_back(EncodeVopc(0xe1, 4, 0));
+  code.push_back(EncodeVop2(0x01, 3, InlineU32(0), 2));
+  AppendStoreVgpr(&code, 3, 0);
+
+  AppendSMovLiteral(&code, 4, 6);
+  code.push_back(EncodeVopc(0xe1, 4, 0));
+  code.push_back(EncodeVop2(0x01, 4, InlineU32(0), 2));
+  AppendStoreVgpr(&code, 4, 1);
+  AppendEnd(&code);
+
+  TestCase test{"VectorVopcCompareLtU64OnGpu",
+                code,
+                {},
+                {1, 0},
+                {O::V_MOV_B32, O::S_MOV_B32, O::V_CMP_LT_U64,
+                 O::V_CNDMASK_B32, O::BUFFER_STORE_DWORD, O::S_ENDPGM}};
+  test.decoded_counts = {{"V_CMP_LT_U64 vcc_lo, s4, v0", 2}};
+  return test;
+}
+
 TestCase VectorVop3CompareEqI64OnGpu() {
   using O = ShaderOpcode;
 
@@ -16646,30 +16707,6 @@ TestCase ScalarMemoryLoadVariants() {
            O::BUFFER_STORE_DWORD, O::S_ENDPGM}};
 }
 
-TestCase ScalarLoadUsesRawExecAddress() {
-  using O = ShaderOpcode;
-
-  std::vector<u32> code;
-  code.push_back(EncodeSMovB32(0, InlineU32(0)));
-  code.push_back(EncodeSMovB32(1, InlineU32(0)));
-  code.push_back(EncodeSMovB32(2, InlineU32(4)));
-  code.push_back(EncodeSMovB32(3, InlineU32(0)));
-  code.push_back(EncodeSop1(0x04, 126, 2));
-  code.push_back(EncodeSmem0(0x00, 20, 63));
-  code.push_back(EncodeSmem1(0));
-  code.push_back(EncodeSMovB32(2, InlineU32(1)));
-  code.push_back(EncodeSop1(0x04, 126, 2));
-  AppendStoreSgpr(&code, 20, 0);
-  AppendEnd(&code);
-
-  return {"ScalarLoadUsesRawExecAddress",
-          code,
-          {0x11111111u, 0x22222222u},
-          {0x22222222u, 0x22222222u},
-          {O::S_MOV_B32, O::S_MOV_B64, O::S_LOAD_DWORD, O::V_MOV_B32,
-           O::BUFFER_STORE_DWORD, O::S_ENDPGM}};
-}
-
 TestCase ScalarLoadSignedImmediateOffsetAddsSoffset() {
   using O = ShaderOpcode;
 
@@ -21113,11 +21150,9 @@ std::vector<TestCase> MakeCases() {
   AddCase(ScalarCompareOps);
   AddCase(ScalarShiftAddAndMaskOps);
   AddCase(ScalarNotB64UpdatesScc);
-  AddCase(ScalarMovB64PreservesRawSpecialData);
-  AddCase(Scalar64RawSpecialOps);
-  AddCase(ScalarMovB32SpecialAliases);
   AddCase(ScalarQuadmaskB64);
   AddCase(ScalarFlbitI32B64Gpu);
+  AddCase(ScalarFlbitI32B32Gpu);
   AddCase(ScalarSaveExecOps);
   AddCase(ScalarOrn2SaveexecUsesSourceOrNotExec);
   AddCase(ScalarGetpcWritesNextInstructionPc);
@@ -21141,6 +21176,7 @@ std::vector<TestCase> MakeCases() {
   AddCase(Vop2SdwaSubNcExactByte2Destination);
   AddCase(Vop2SdwaAddNcCapturedHighWordDestination);
   AddCase(Vop2SdwaAshrrevCapturedWord0SignExtends);
+  AddCase(Vop2SdwaLshrrevCapturedByte1Source);
   AddCase(Vop2SdwaSubNcPreservesByteAndWordDestinations);
   AddCase(Vop3CvtPkI16I32Captured);
   AddCase(Vop3Med3I16Captured);
@@ -21183,6 +21219,7 @@ std::vector<TestCase> MakeCases() {
   AddCase(VectorPermlane16FetchInactiveZero);
   AddCase(VectorPermlane16FetchInactiveFi);
   AddCase(VectorDppQuadPermuteReverse);
+  AddCase(VectorDppRowXmask);
   AddCase(VectorDppBankMaskPreservesDestination);
   AddCase(VectorDppBoundsControlZeroPreservesDestination);
   AddCase(Vop3LdexpSourceModifier);
@@ -21207,6 +21244,7 @@ std::vector<TestCase> MakeCases() {
   AddCase(VectorVop3CompareEqI64OnGpu);
   AddCase(VectorVop3CompareEqU64OnGpu);
   AddCase(VectorVop3CompareGtU64OnGpu);
+  AddCase(VectorVopcCompareLtU64OnGpu);
   AddCase(VectorVop3CompareNeU64OnGpu);
   AddCase(VectorVopcCmpxNeU64CapturedExecMask);
   AddCase(VectorVop3CmpxNeI64CapturedExecMask);
@@ -21229,7 +21267,6 @@ std::vector<TestCase> MakeCases() {
   AddCase(BranchVccnzUsesInvocationMask);
   AddCase(BranchVccnzUsesCarryProducedInvocationMask);
   AddCase(ScalarMemoryLoadVariants);
-  AddCase(ScalarLoadUsesRawExecAddress);
   AddCase(ScalarLoadSignedImmediateOffsetAddsSoffset);
   AddCase(ScalarLoadAlignsComponentsAndMasksAddress);
   AddCase(BufferLoadStore);
@@ -21425,8 +21462,7 @@ void CheckPs5GameExampleImageClearRuntimeShape() {
     options.stage = ShaderType::Compute;
     options.wave_size = 64;
     options.user_data_base = 0;
-    options.user_data_count = static_cast<u32>(user_data.size());
-    options.user_data = user_data.data();
+    options.user_data = user_data;
     options.input_info.compute = &compute;
     options.dump_ir = false;
     auto result = ShaderRecompiler::Recompile(code, options);
@@ -21522,8 +21558,7 @@ void CheckEmbeddedFetchVertexOffset() {
     ShaderRecompiler::CompileOptions options;
     options.stage = ShaderType::Vertex;
     options.user_data_base = 8;
-    options.user_data_count = static_cast<u32>(user_data.size());
-    options.user_data = user_data.data();
+    options.user_data = user_data;
     options.input_info.vertex = &vertex;
 
     auto result = ShaderRecompiler::Recompile(code, options);
@@ -23771,18 +23806,18 @@ void CheckNativeMsaaState() {
 }
 
 void CheckDepthHtileStencilCompatibility() {
+  Require("DepthHtileStencilCompatibility", "inactive stencil plane",
+          depth_htile_stencil_acceleration_compatible(false, true, false),
+          "inactive Hi-Stencil state was treated as an attachment");
   Require("DepthHtileStencilCompatibility", "disabled acceleration",
-          depth_htile_stencil_acceleration_compatible(false, false, true),
+          depth_htile_stencil_acceleration_compatible(true, false, true),
           "disabled Hi-Stencil state was rejected");
   Require("DepthHtileStencilCompatibility", "PS5 stencil plus HTile",
           depth_htile_stencil_acceleration_compatible(true, true, false),
           "valid PS5 Hi-Stencil attachment was rejected");
-  Require("DepthHtileStencilCompatibility", "missing stencil plane",
-          !depth_htile_stencil_acceleration_compatible(false, true, false),
-          "Hi-Stencil without a stencil plane was silently admitted");
   Require("DepthHtileStencilCompatibility", "missing HTile metadata",
           !depth_htile_stencil_acceleration_compatible(true, false, false),
-          "Hi-Stencil without HTile metadata was silently admitted");
+          "active Hi-Stencil without HTile metadata was admitted");
   std::printf("[host]    %-32s ok\n", "DepthHtileStencilCompatibility");
 }
 
@@ -24223,8 +24258,7 @@ void CheckEmbeddedFetchLaneSpill() {
   ShaderRecompiler::CompileOptions options;
   options.stage = ShaderType::Vertex;
   options.user_data_base = 8;
-  options.user_data_count = static_cast<u32>(user_data.size());
-  options.user_data = user_data.data();
+  options.user_data = user_data;
   options.input_info.vertex = &vertex;
 
   auto result = ShaderRecompiler::Recompile(code, options);
@@ -24370,6 +24404,189 @@ void CheckPm4StencilInfoValueLane(RenderContext &renderer) {
   std::printf("[host]    %-32s ok\n", "Pm4StencilInfoValueLane");
 }
 
+void CheckPm4NativeTargetGeometryRegisters(RenderContext &renderer) {
+  GraphicsInitJmpTables();
+  CommandProcessor packet(renderer, 0);
+  CommandProcessor indirect(renderer, 0);
+
+  const auto write_context = [](CommandProcessor &processor, uint32_t offset,
+                                uint32_t value) {
+    std::array<uint32_t, 3> command{
+        KYTY_PM4(3, Pm4::IT_SET_CONTEXT_REG, Pm4::R_ZERO), offset, value};
+    Pm4Execution execution;
+    return processor.Process(execution, command) == Pm4ProcessResult::Complete;
+  };
+
+  constexpr uint32_t color_slot = 7;
+  constexpr uint32_t width_minus_one = 1919;
+  constexpr uint32_t height_minus_one = 1079;
+  constexpr uint32_t depth_minus_one = 5;
+  constexpr uint32_t max_mip = 3;
+  constexpr uint32_t dimension_3d = 2;
+  constexpr uint32_t attrib2_value =
+      height_minus_one | (width_minus_one << 14u) | (max_mip << 28u);
+  constexpr uint32_t attrib3_value =
+      depth_minus_one |
+      (static_cast<uint32_t>(Prospero::TileMode::kStandard64KB) << 14u) |
+      (dimension_3d << 24u) | (1u << 26u) | (1u << 30u);
+
+  const bool color_packets_complete =
+      write_context(packet, Pm4::CB_COLOR7_ATTRIB2, attrib2_value) &&
+      write_context(packet, Pm4::CB_COLOR7_ATTRIB3, attrib3_value);
+  g_hw_ctx_indirect_func[Pm4::CB_COLOR7_ATTRIB2](
+      indirect, Pm4::CB_COLOR7_ATTRIB2, attrib2_value);
+  g_hw_ctx_indirect_func[Pm4::CB_COLOR7_ATTRIB3](
+      indirect, Pm4::CB_COLOR7_ATTRIB3, attrib3_value);
+
+  const auto color_matches = [](const HW::RenderTarget &target) {
+    return target.attrib2.width == width_minus_one &&
+           target.attrib2.height == height_minus_one &&
+           target.attrib2.num_mip_levels == max_mip &&
+           target.attrib3.depth == depth_minus_one &&
+           target.attrib3.tile_mode == Prospero::TileMode::kStandard64KB &&
+           target.attrib3.dimension == dimension_3d &&
+           target.attrib3.metadata_pipe_aligned;
+  };
+  Require("Pm4NativeTargetGeometry", "sparse color geometry",
+          color_packets_complete &&
+              color_matches(packet.GetCtx().GetRenderTarget(color_slot)) &&
+              color_matches(indirect.GetCtx().GetRenderTarget(color_slot)) &&
+              !color_matches(packet.GetCtx().GetRenderTarget(0)) &&
+              !color_matches(indirect.GetCtx().GetRenderTarget(0)),
+          "ATTRIB2/ATTRIB3 did not provide native color extent and layout");
+
+  constexpr uint32_t depth_size_value =
+      width_minus_one | (height_minus_one << 16u);
+  const bool depth_packet_complete =
+      write_context(packet, Pm4::DB_DEPTH_SIZE_XY, depth_size_value);
+  g_hw_ctx_indirect_func[Pm4::DB_DEPTH_SIZE_XY](indirect, Pm4::DB_DEPTH_SIZE_XY,
+                                                depth_size_value);
+  const auto depth_matches = [](const HW::DepthRenderTarget &target) {
+    return target.size.valid && target.size.x_max == width_minus_one &&
+           target.size.y_max == height_minus_one;
+  };
+  Require("Pm4NativeTargetGeometry", "native depth geometry",
+          depth_packet_complete &&
+              depth_matches(packet.GetCtx().GetDepthRenderTarget()) &&
+              depth_matches(indirect.GetCtx().GetDepthRenderTarget()),
+          "DB_DEPTH_SIZE_XY did not provide the native depth extent");
+
+  bool legacy_slots_are_unhandled = true;
+  for (uint32_t slot = 0; slot < 8; slot++) {
+    const auto stride = slot * 15u;
+    const std::array legacy_color_slots{
+        Pm4::CB_COLOR0_BASE + stride + 1u,
+        Pm4::CB_COLOR0_BASE + stride + 2u,
+        Pm4::CB_COLOR0_CMASK + stride + 1u,
+        Pm4::CB_COLOR0_FMASK + stride + 1u,
+    };
+    for (const auto offset : legacy_color_slots) {
+      legacy_slots_are_unhandled &= g_hw_ctx_func[offset] == nullptr &&
+                                    g_hw_ctx_indirect_func[offset] == nullptr;
+    }
+  }
+  // These offsets are the removed GCN DB_DEPTH_INFO, DB_DEPTH_SIZE, and
+  // DB_DEPTH_SLICE holes. Keep them numeric so deleting their legacy PM4 names
+  // does not weaken this dispatch-table regression check.
+  constexpr std::array<uint32_t, 3> legacy_depth_slots{0x000fu, 0x0016u,
+                                                       0x0017u};
+  for (const auto offset : legacy_depth_slots) {
+    legacy_slots_are_unhandled &= g_hw_ctx_func[offset] == nullptr &&
+                                  g_hw_ctx_indirect_func[offset] == nullptr;
+  }
+  // Removed GCN PA_SC_NGG_MODE_CNTL.
+  legacy_slots_are_unhandled &= g_hw_ctx_func[0x0314u] == nullptr &&
+                                g_hw_ctx_indirect_func[0x0314u] == nullptr;
+
+  // Removed GCN shader resource/checksum/queue registers. Numeric offsets keep
+  // this check independent of the deleted legacy names.
+  constexpr std::array<uint32_t, 19> legacy_shader_slots{
+      0x000u, 0x001u, 0x002u, 0x003u, 0x030u, 0x0b0u, 0x0bcu, 0x130u,
+      0x14au, 0x14bu, 0x20eu, 0x20fu, 0x210u, 0x211u, 0x216u, 0x217u,
+      0x219u, 0x21au, 0x27du,
+  };
+  for (const auto offset : legacy_shader_slots) {
+    legacy_slots_are_unhandled &= g_hw_sh_func[offset] == nullptr &&
+                                  g_hw_sh_indirect_func[offset] == nullptr;
+  }
+  for (uint32_t offset = 0x0cau; offset <= 0x0ebu; offset++) {
+    legacy_slots_are_unhandled &= g_hw_sh_func[offset] == nullptr &&
+                                  g_hw_sh_indirect_func[offset] == nullptr;
+  }
+
+  std::array<uint32_t, 3> index_size_packet{
+      KYTY_PM4(3, Pm4::IT_SET_UCONFIG_REG_INDEX, Pm4::R_ZERO),
+      0x20000000u | Pm4::VGT_INDEX_TYPE,
+      0x441u,
+  };
+  Pm4Execution index_size_execution;
+  const bool native_index_size_packet_complete =
+      packet.Process(index_size_execution, index_size_packet) ==
+          Pm4ProcessResult::Complete &&
+      g_hw_uc_func[Pm4::VGT_INDEX_TYPE] != nullptr &&
+      g_hw_uc_indirect_func[Pm4::VGT_INDEX_TYPE] != nullptr &&
+      g_hw_uc_func[Pm4::IA_MULTI_VGT_PARAM] != nullptr &&
+      g_hw_uc_indirect_func[Pm4::IA_MULTI_VGT_PARAM] != nullptr;
+
+  const bool native_pace_slots_are_handled =
+      g_hw_sh_indirect_func[Pm4::SPI_SHADER_PACE_ID_PS] != nullptr &&
+      g_hw_sh_indirect_func[Pm4::SPI_SHADER_PACE_ID_GS] != nullptr &&
+      g_hw_sh_indirect_func[Pm4::COMPUTE_PACE_ID] != nullptr;
+  Require("Pm4NativeTargetGeometry", "legacy register holes",
+          legacy_slots_are_unhandled && native_pace_slots_are_handled &&
+              native_index_size_packet_complete,
+          "a removed GCN register has a handler or a native packet is not handled");
+  std::printf("[host]    %-32s ok\n", "Pm4NativeTargetGeometry");
+}
+
+void CheckPm4PrivateAgcShaderRegisters(RenderContext &renderer) {
+  GraphicsInitJmpTables();
+  CommandProcessor processor(renderer, 0);
+  std::array<uint32_t, 6> registers{
+      Pm4::SPI_SHADER_PGM_RSRC4_GS, 0x0badc0deu,
+      Pm4::SPI_SHADER_PGM_CHKSUM_HS, 0x12345678u,
+      Pm4::SPI_SHADER_PGM_RSRC4_HS, 0x87654321u,
+  };
+  const auto address = reinterpret_cast<uint64_t>(registers.data());
+  std::array<uint32_t, 5> command{
+      KYTY_PM4(5, Pm4::IT_SET_SH_REG_INDIRECT, Pm4::R_ZERO),
+      static_cast<uint32_t>(address), static_cast<uint32_t>(address >> 32u),
+      0x80000000u, 3u,
+  };
+  Pm4Execution execution;
+  const bool handlers_present =
+      g_hw_sh_indirect_func[Pm4::SPI_SHADER_PGM_RSRC4_GS] != nullptr &&
+      g_hw_sh_indirect_func[Pm4::SPI_SHADER_PGM_CHKSUM_HS] != nullptr &&
+      g_hw_sh_indirect_func[Pm4::SPI_SHADER_PGM_RSRC4_HS] != nullptr;
+  Require("Pm4PrivateAgcShaderRegisters", "private shader register stream",
+          handlers_present &&
+              processor.Process(execution, command) == Pm4ProcessResult::Complete,
+          "private AGC shader registers were rejected by SET_SH_REG_INDIRECT");
+  std::printf("[host]    %-32s ok\n", "Pm4PrivateAgcShaderRegisters");
+}
+
+void CheckPm4PrivateAgcUconfigRegisters(RenderContext &renderer) {
+  GraphicsInitJmpTables();
+  CommandProcessor processor(renderer, 0);
+  std::array<uint32_t, 2> registers{
+      Pm4::UC_PARAMETER_OVERSUBSCRIPTION, 0x12345678u,
+  };
+  const auto address = reinterpret_cast<uint64_t>(registers.data());
+  std::array<uint32_t, 5> command{
+      KYTY_PM4(5, Pm4::IT_SET_UCONFIG_REG_INDIRECT, Pm4::R_ZERO),
+      static_cast<uint32_t>(address), static_cast<uint32_t>(address >> 32u),
+      0x80000000u, 1u,
+  };
+  Pm4Execution execution;
+  const bool handler_present =
+      g_hw_uc_indirect_func[Pm4::UC_PARAMETER_OVERSUBSCRIPTION] != nullptr;
+  Require("Pm4PrivateAgcUconfigRegisters", "private UCONFIG register stream",
+          handler_present &&
+              processor.Process(execution, command) == Pm4ProcessResult::Complete,
+          "private AGC UCONFIG register was rejected by SET_UCONFIG_REG_INDIRECT");
+  std::printf("[host]    %-32s ok\n", "Pm4PrivateAgcUconfigRegisters");
+}
+
 void CheckPm4DirectShaderRegisterFallback(RenderContext &renderer) {
   GraphicsInitJmpTables();
   CommandProcessor processor(renderer, 0);
@@ -24532,6 +24749,54 @@ void CheckPm4PolygonOffsetRegisters(RenderContext &renderer) {
   std::printf("[host]    %-32s ok\n", "Pm4PolygonOffset");
 }
 
+void CheckPm4DepthControlHighBits(RenderContext &renderer) {
+  GraphicsInitJmpTables();
+  CommandProcessor processor(renderer, 0);
+
+  std::array<uint32_t, 3> direct{
+      KYTY_PM4(3, Pm4::IT_SET_CONTEXT_REG, Pm4::R_ZERO),
+      Pm4::DB_DEPTH_CONTROL,
+      0x80000072u,
+  };
+  Pm4Execution direct_execution;
+  const bool direct_complete =
+      processor.Process(direct_execution, direct) == Pm4ProcessResult::Complete;
+  const auto &direct_control = processor.GetCtx().GetDepthControl();
+  const bool direct_decoded =
+      !direct_control.stencil_enable && direct_control.z_enable &&
+      !direct_control.z_write_enable && !direct_control.depth_bounds_enable &&
+      direct_control.zfunc == 7u && !direct_control.backface_enable &&
+      direct_control.stencilfunc == 0u && direct_control.stencilfunc_bf == 0u;
+
+  std::array<uint32_t, 2> registers{
+      Pm4::DB_DEPTH_CONTROL,
+      0x4070078du,
+  };
+  const auto address = reinterpret_cast<uint64_t>(registers.data());
+  std::array<uint32_t, 5> indirect{
+      KYTY_PM4(5, Pm4::IT_SET_CONTEXT_REG_INDIRECT, Pm4::R_ZERO),
+      static_cast<uint32_t>(address), static_cast<uint32_t>(address >> 32u),
+      0x80000000u, 1u,
+  };
+  Pm4Execution indirect_execution;
+  const bool indirect_complete =
+      processor.Process(indirect_execution, indirect) ==
+      Pm4ProcessResult::Complete;
+  const auto &indirect_control = processor.GetCtx().GetDepthControl();
+  const bool indirect_decoded =
+      indirect_control.stencil_enable && !indirect_control.z_enable &&
+      indirect_control.z_write_enable && indirect_control.depth_bounds_enable &&
+      indirect_control.zfunc == 0u && indirect_control.backface_enable &&
+      indirect_control.stencilfunc == 7u &&
+      indirect_control.stencilfunc_bf == 7u;
+
+  Require("Pm4DepthControlHighBits", "preserved high bits",
+          direct_complete && direct_decoded && indirect_complete &&
+              indirect_decoded,
+          "DB_DEPTH_CONTROL high bits rejected or changed decoded PS5 state");
+  std::printf("[host]    %-32s ok\n", "Pm4DepthControlHighBits");
+}
+
 struct AgcCommandBufferLayout {
   using Callback = KYTY_SYSV_ABI bool (*)(Gen5::CommandBuffer *, uint32_t,
                                           void *);
@@ -25129,8 +25394,12 @@ int main(int argc, char **argv) {
   }
   if (argc == 2 && std::strcmp(argv[1], "--context-state-only") == 0) {
     VulkanHarness vulkan;
+    CheckPm4NativeTargetGeometryRegisters(vulkan.RuntimeRenderer());
+    CheckPm4PrivateAgcShaderRegisters(vulkan.RuntimeRenderer());
+    CheckPm4PrivateAgcUconfigRegisters(vulkan.RuntimeRenderer());
     CheckPm4BlendColorRegisterRanges(vulkan.RuntimeRenderer());
     CheckPm4PolygonOffsetRegisters(vulkan.RuntimeRenderer());
+    CheckPm4DepthControlHighBits(vulkan.RuntimeRenderer());
     CheckPm4ContextStateOperations(vulkan.RuntimeRenderer());
     return 0;
   }
@@ -25167,6 +25436,11 @@ int main(int argc, char **argv) {
     vulkan.CheckUnifiedTextureCacheFlow();
     return 0;
   }
+  if (argc == 2 && std::strcmp(argv[1], "--compute-meta-clear-only") == 0) {
+    VulkanHarness vulkan;
+    vulkan.CheckComputeMetaClearClassification();
+    return 0;
+  }
   if (argc == 2 && std::strcmp(argv[1], "--layered-image-only") == 0) {
     VulkanHarness vulkan;
     vulkan.CheckUnifiedImageViewCache();
@@ -25308,10 +25582,14 @@ int main(int argc, char **argv) {
   CheckPm4AcquireMemNoOp(vulkan.RuntimeRenderer());
   CheckPm4SyntheticOcclusionCounterDump(vulkan.RuntimeRenderer());
   CheckPm4StencilInfoValueLane(vulkan.RuntimeRenderer());
+  CheckPm4NativeTargetGeometryRegisters(vulkan.RuntimeRenderer());
+  CheckPm4PrivateAgcShaderRegisters(vulkan.RuntimeRenderer());
+  CheckPm4PrivateAgcUconfigRegisters(vulkan.RuntimeRenderer());
   CheckPm4DirectShaderRegisterFallback(vulkan.RuntimeRenderer());
   CheckPm4GuardBandRegisterRanges(vulkan.RuntimeRenderer());
   CheckPm4BlendColorRegisterRanges(vulkan.RuntimeRenderer());
   CheckPm4PolygonOffsetRegisters(vulkan.RuntimeRenderer());
+  CheckPm4DepthControlHighBits(vulkan.RuntimeRenderer());
   CheckAgcWaitPackets(vulkan.RuntimeRenderer());
   CheckAgcDrawIndirectMultiPacket(vulkan.RuntimeRenderer());
   CheckPm4ContextStateOperations(vulkan.RuntimeRenderer());

tests/shaderCfgTests.cpp

@@ -28,6 +28,7 @@
 #include "graphics/shader/shaderCompiler.h"
 #include "libs/agc.h"
 #include "spirv-tools/libspirv.hpp"
+#include "xxhash.h"
 
 #include <algorithm>
 #include <array>
@@ -84,9 +85,11 @@ ShaderRecompiler::CompileOptions MakeCompileOptions(ShaderType stage) {
   static const ShaderVertexInputInfo vertex{};
   static const ShaderPixelInputInfo pixel{};
   static const ShaderComputeInputInfo compute{};
+  static const std::array<uint32_t, 64> user_data{};
 
   ShaderRecompiler::CompileOptions options;
   options.stage = stage;
+  options.user_data = user_data;
   switch (stage) {
   case ShaderType::Vertex:
     options.input_info.vertex = &vertex;
@@ -115,9 +118,7 @@ void CompilePixelRuntime(const ShaderParams &params,
                          ShaderPixelInputInfo &input_info) {
   auto options = MakeCompileOptions(ShaderType::Pixel);
   options.shader_hash = params.hash;
-  options.shader_base = params.Base();
-  options.user_data = params.user_data.data();
-  options.user_data_count = static_cast<uint32_t>(params.user_data.size());
+  options.user_data = params.user_data;
   options.scratch_dwords = input_info.scratch_size_dwords;
   options.push_constant_offset = input_info.push_constant_offset;
   options.read_specialization_memory = ReadHostTestMemory;
@@ -3696,7 +3697,6 @@ void TestNewShaderRecompilerScalarB64LaneTranslation() {
   const uint32_t user_data[] = {0x01234567u, 0x89abcdefu};
   auto options = MakeCompileOptions(ShaderType::Vertex);
   options.user_data = user_data;
-  options.user_data_count = static_cast<uint32_t>(std::size(user_data));
   options.dump_ir = true;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
@@ -3969,6 +3969,37 @@ void TestNewShaderDecoderArchitecture() {
             cmp_eq_u64.src1.kind == OperandKind::VccLo,
         "decoder rejected the captured VOP3 V_CMP_EQ_U64 instruction");
 
+  const uint32_t completed_pr_opcodes[] = {
+      EncodeSop1(0x15, 0, 1),   // s_flbit_i32_b32 s0, s1
+      EncodeVopc(0xe1, 128, 0), // v_cmp_lt_u64 vcc, 0, v[0:1]
+      EncodeSopp(0x01, 0),
+  };
+  Instruction flbit_i32_b32;
+  ShaderRecompiler::Decoder::DecodeInstruction(completed_pr_opcodes, 0u,
+                                               flbit_i32_b32);
+  Instruction cmp_lt_u64;
+  ShaderRecompiler::Decoder::DecodeInstruction(completed_pr_opcodes, 1u,
+                                               cmp_lt_u64);
+  Check(flbit_i32_b32.family == Family::SOP1 &&
+            flbit_i32_b32.opcode == Opcode::S_FLBIT_I32_B32 &&
+            flbit_i32_b32.dst.kind == OperandKind::Sgpr &&
+            flbit_i32_b32.dst.reg == 0u &&
+            flbit_i32_b32.src0.kind == OperandKind::Sgpr &&
+            flbit_i32_b32.src0.reg == 1u &&
+            cmp_lt_u64.family == Family::VOPC &&
+            cmp_lt_u64.opcode == Opcode::V_CMP_LT_U64 &&
+            cmp_lt_u64.dst.kind == OperandKind::VccLo,
+        "decoder rejected completed PR #427 scalar/vector opcodes");
+
+  const auto completed_options = MakeCompileOptions(ShaderType::Compute);
+  const auto completed_result =
+      ShaderRecompiler::Recompile(completed_pr_opcodes, completed_options);
+  Check(Common::ContainsStr(completed_result.decoded_dump,
+                            "S_FLBIT_I32_B32 s0, s1") &&
+            Common::ContainsStr(completed_result.decoded_dump,
+                                "V_CMP_LT_U64 vcc_lo, 0, v0"),
+        "completed PR #427 opcodes were absent from the decoded shader");
+
   const uint32_t ffbh_i32_code[] = {0x7e1c770eu};
   Instruction ffbh_i32;
   ShaderRecompiler::Decoder::DecodeInstruction(ffbh_i32_code, 0u, ffbh_i32);
@@ -4369,7 +4400,7 @@ void TestNewShaderRecompilerMemoryFamilyTranslation() {
   auto user_data = ImageTestUserData();
   auto options = MakeCompileOptions(ShaderType::Compute);
   options.dump_ir = true;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
   options.read_memory = ReadZeroTestMemory;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
@@ -4463,8 +4494,7 @@ void TestNewShaderRecompilerScalarMemoryBindingDomains() {
   raw_user_data[8] = 0x1003u;
 
   auto raw_options = MakeCompileOptions(ShaderType::Pixel);
-  raw_options.user_data = raw_user_data.data();
-  raw_options.user_data_count = static_cast<uint32_t>(raw_user_data.size());
+  raw_options.user_data = raw_user_data;
 
   auto raw = ShaderRecompiler::Recompile(raw_shader, raw_options);
   const auto *address_binding = ShaderRecompiler::IR::FindBinding(
@@ -4508,9 +4538,7 @@ void TestNewShaderRecompilerScalarMemoryBindingDomains() {
   buffer_user_data[10] = 4u;
 
   auto buffer_options = MakeCompileOptions(ShaderType::Pixel);
-  buffer_options.user_data = buffer_user_data.data();
-  buffer_options.user_data_count =
-      static_cast<uint32_t>(buffer_user_data.size());
+  buffer_options.user_data = buffer_user_data;
 
   auto buffer = ShaderRecompiler::Recompile(buffer_shader, buffer_options);
   const auto *buffer_binding = ShaderRecompiler::IR::FindBinding(
@@ -4548,7 +4576,7 @@ void TestNewShaderRecompilerImageQueryTranslation() {
   auto user_data = ImageTestUserData();
   auto options = MakeCompileOptions(ShaderType::Compute);
   options.dump_ir = true;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   Check(Common::ContainsStr(result.decoded_dump, "image_get_lod"),
@@ -4592,7 +4620,7 @@ void TestNewShaderRecompilerCubeSampleCoordinates() {
 
   auto user_data = ImageTestUserData(Prospero::ImageType::kCube);
   auto options = MakeCompileOptions(ShaderType::Compute);
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   Check(result.program.info.images.size() == 1 &&
@@ -4638,7 +4666,7 @@ void TestNewShaderRecompilerImageSampleVariants() {
   auto user_data = ImageTestUserData();
   auto options = MakeCompileOptions(ShaderType::Compute);
   options.dump_ir = true;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   Check(Common::ContainsStr(result.decoded_dump, "image_sample_l"),
@@ -4726,7 +4754,7 @@ void TestNewShaderRecompilerImageSampleA16SamplerCoords() {
   auto user_data = ImageTestUserData(Prospero::ImageType::kColor3D);
   auto options = MakeCompileOptions(ShaderType::Compute);
   options.dump_ir = true;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   Check(Common::ContainsStr(result.decoded_dump,
@@ -4768,7 +4796,7 @@ void TestNewShaderRecompilerImageSampleOpcodeAliases() {
   auto user_data = ImageTestUserData();
   auto options = MakeCompileOptions(ShaderType::Compute);
   options.dump_ir = true;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   Check(Common::ContainsStr(result.decoded_dump, "image_sample_a"),
@@ -4804,7 +4832,7 @@ void TestNewShaderRecompilerImageSampleA16ExceptionComponents() {
     auto user_data = ImageTestUserData();
     auto options = MakeCompileOptions(ShaderType::Compute);
     options.dump_ir = true;
-    options.user_data = user_data.data();
+    options.user_data = user_data;
 
     auto result = ShaderRecompiler::Recompile(shader, options);
     CheckSpirvBinaryValidates(result.spirv);
@@ -4850,7 +4878,7 @@ void TestNewShaderRecompilerImageLoadA16UintCoords() {
   auto user_data = ImageTestUserData();
   auto options = MakeCompileOptions(ShaderType::Compute);
   options.dump_ir = true;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   Check(Common::ContainsStr(result.decoded_dump, "image_load"),
@@ -4894,7 +4922,7 @@ void TestNewShaderRecompilerPixelImageSampleLodSelection() {
     auto options = MakeCompileOptions(ShaderType::Pixel);
     options.input_info.pixel = &ps_info;
     options.dump_ir = true;
-    options.user_data = user_data.data();
+    options.user_data = user_data;
 
     auto result = ShaderRecompiler::Recompile(shader, options);
     CheckSpirvBinaryValidates(result.spirv);
@@ -5005,7 +5033,7 @@ void TestNewShaderRecompilerImageViewDimensions() {
   auto options = MakeCompileOptions(ShaderType::Pixel);
   options.input_info.pixel = &ps_info;
   options.dump_ir = true;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   Check(Common::ContainsStr(result.decoded_dump, "image_dim=2d_array"),
@@ -5082,7 +5110,7 @@ void TestNewShaderRecompilerStorageImage1DDescriptorVariants() {
 
   auto options = MakeCompileOptions(ShaderType::Compute);
   options.dump_ir = true;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   Check(Common::ContainsStr(result.ir_dump, "image_dim=1d"),
@@ -5113,7 +5141,7 @@ void TestNewShaderRecompilerNullImageUsesCanonical2DView() {
   std::array<uint32_t, 64> user_data{};
 
   auto options = MakeCompileOptions(ShaderType::Compute);
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   CheckSpirvBinaryValidates(result.spirv);
@@ -5144,7 +5172,7 @@ void TestNewShaderRecompilerImageGatherVariants() {
   auto user_data = ImageTestUserData();
   auto options = MakeCompileOptions(ShaderType::Compute);
   options.dump_ir = true;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   Check(Common::ContainsStr(result.decoded_dump, "image_gather4_lz"),
@@ -5230,7 +5258,7 @@ void TestNewShaderRecompilerImageLoadVariants() {
   auto user_data = ImageTestUserData();
   auto options = MakeCompileOptions(ShaderType::Compute);
   options.dump_ir = true;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   Check(Common::ContainsStr(result.decoded_dump, "image_load"),
@@ -5276,7 +5304,7 @@ void TestNewShaderRecompilerImageLoad2DMsaa() {
 
   auto options = MakeCompileOptions(ShaderType::Pixel);
   options.dump_ir = true;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   Check(Common::ContainsStr(result.decoded_dump, "image_dim=2d_msaa") &&
@@ -5317,7 +5345,7 @@ void TestNewShaderRecompilerImageStoreTranslation() {
   auto user_data = ImageTestUserData();
   auto options = MakeCompileOptions(ShaderType::Compute);
   options.dump_ir = true;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   Check(Common::ContainsStr(result.decoded_dump, "image_store"),
@@ -5368,7 +5396,7 @@ void TestNewShaderRecompilerStorageImage3DDescriptorVariant() {
   auto options = MakeCompileOptions(ShaderType::Compute);
   options.dump_ir = true;
   options.input_info.compute = &input_info;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   Check(Common::ContainsStr(result.decoded_dump, "image_dim=3d"),
@@ -5422,7 +5450,7 @@ void TestNewShaderRecompilerStorageImage2DDescriptorOverridesMimg3D() {
   auto options = MakeCompileOptions(ShaderType::Compute);
   options.dump_ir = true;
   options.input_info.compute = &input_info;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   Check(Common::ContainsStr(result.decoded_dump, "image_dim=3d"),
@@ -5464,7 +5492,7 @@ void TestNewShaderRecompilerImageAtomicTranslation() {
   auto user_data = ImageTestUserData();
   auto options = MakeCompileOptions(ShaderType::Compute);
   options.dump_ir = true;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   Check(Common::ContainsStr(result.decoded_dump, "image_atomic_add"),
@@ -6127,7 +6155,7 @@ void TestNewShaderRecompilerFormattedStoreUsesRuntimeArrayLengthOnly() {
 
   auto options = MakeCompileOptions(ShaderType::Compute);
   options.dump_ir = true;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   Check(result.resources.buffers.size() == 1 &&
@@ -8783,6 +8811,45 @@ void TestNewShaderRecompilerExpVertexOutputs() {
   CheckSpirvBinaryValidates(result.spirv);
 }
 
+void TestNewShaderRecompilerClipDisabledPosition() {
+  const uint32_t shader[] = {
+      EncodeExp0(0x0c, 0xf), EncodeExp1(0, 1, 2, 3), 0xbf810000u,
+  };
+  const auto compile = [&](const ShaderVertexInputInfo &vertex) {
+    auto options = MakeCompileOptions(ShaderType::Vertex);
+    options.input_info.vertex = &vertex;
+    return ShaderRecompiler::Recompile(shader, options);
+  };
+
+  ShaderVertexInputInfo regular{};
+  auto clipped = regular;
+  clipped.clip_space = {
+      .scale = {640.0f, 360.0f},
+      .offset = {640.0f, 360.0f},
+      .half_extent = {8192.0f, 8192.0f},
+      .enabled = true,
+  };
+  const auto regular_result = compile(regular);
+  const auto clipped_result = compile(clipped);
+  CheckSpirvBinaryValidates(clipped_result.spirv);
+  Check(SpirvInstructionOpcodeCount(regular_result.spirv, 136u) == 0u &&
+            SpirvInstructionOpcodeCount(clipped_result.spirv, 133u) == 2u &&
+            SpirvInstructionOpcodeCount(clipped_result.spirv, 129u) == 2u &&
+            SpirvInstructionOpcodeCount(clipped_result.spirv, 136u) == 2u &&
+            SpirvInstructionOpcodeCount(clipped_result.spirv, 131u) == 2u,
+        "clip-disabled position export did not convert both screen coordinates to NDC");
+  Check(MakeStageStaticKey(regular) != MakeStageStaticKey(clipped),
+        "clip-disabled vertex shader shared the regular shader cache key");
+  auto disabled = clipped;
+  disabled.clip_space.enabled = false;
+  Check(MakeStageStaticKey(regular) == MakeStageStaticKey(disabled),
+        "disabled clip-space payload affected the shader cache key");
+  auto shifted = clipped;
+  shifted.clip_space.offset[0] += 1.0f;
+  Check(MakeStageStaticKey(clipped) != MakeStageStaticKey(shifted),
+        "clip-disabled viewport transform is absent from the shader cache key");
+}
+
 void TestNewShaderRecompilerAuxPositionExports() {
   const auto compile = [](std::span<const uint32_t> shader, uint32_t control) {
     ShaderVertexInputInfo vertex{};
@@ -8891,9 +8958,6 @@ void TestNewShaderRecompilerAuxPositionExports() {
           "compilation");
 #endif
 
-  HW::VertexShaderInfo regs{};
-  regs.es_regs.data_addr = 1;
-  regs.gs_regs.chksum = 1;
   ShaderVertexInputInfo key0{};
   ShaderVertexInputInfo key1{};
   key1.pa_cl_vs_out_cntl = 1;
@@ -8946,7 +9010,7 @@ void TestDemandDrivenSpirvDeclarations() {
     auto user_data = ImageTestUserData();
     SetImageTestFormat(&user_data, 0, format);
     auto options = MakeCompileOptions(ShaderType::Compute);
-    options.user_data = user_data.data();
+    options.user_data = user_data;
     auto result = ShaderRecompiler::Recompile(shader, options);
     CheckSpirvBinaryValidates(result.spirv);
     return result.spirv;
@@ -8959,7 +9023,7 @@ void TestDemandDrivenSpirvDeclarations() {
   auto query_pixel_info = RegressionPixelInputInfo();
   auto query_options = MakeCompileOptions(ShaderType::Pixel);
   query_options.input_info.pixel = &query_pixel_info;
-  query_options.user_data = query_user_data.data();
+  query_options.user_data = query_user_data;
   auto query_result = ShaderRecompiler::Recompile(query, query_options);
   CheckSpirvBinaryValidates(query_result.spirv);
   const auto &query_spirv = query_result.spirv;
@@ -9034,12 +9098,10 @@ void TestTypedEntryStateIsMinimal() {
                                                           translate_options);
 
     uint32_t set_exec = 0;
-    uint32_t set_exec_mask_tag = 0;
     uint32_t set_exec_lo = 0;
     uint32_t set_exec_hi = 0;
     uint32_t ballots = 0;
     IR::Value exec;
-    IR::Value exec_mask_tag;
     IR::Value exec_lo;
     IR::Value exec_hi;
     for (const auto *block : values.blocks) {
@@ -9052,10 +9114,6 @@ void TestTypedEntryStateIsMinimal() {
           set_exec++;
           exec = inst.Arg(0).Resolve();
           break;
-        case IR::ValueOpcode::SetExecMaskTag:
-          set_exec_mask_tag++;
-          exec_mask_tag = inst.Arg(0).Resolve();
-          break;
         case IR::ValueOpcode::SetExecLo:
           set_exec_lo++;
           exec_lo = inst.Arg(0).Resolve();
@@ -9070,7 +9128,6 @@ void TestTypedEntryStateIsMinimal() {
         case IR::ValueOpcode::SetVectorRegister:
         case IR::ValueOpcode::SetScc:
         case IR::ValueOpcode::SetVcc:
-        case IR::ValueOpcode::SetVccMaskValidTag:
         case IR::ValueOpcode::SetVccLo:
         case IR::ValueOpcode::SetVccHi:
         case IR::ValueOpcode::SetM0:
@@ -9081,9 +9138,8 @@ void TestTypedEntryStateIsMinimal() {
         }
       }
     }
-    Check(set_exec == 1u && set_exec_mask_tag == 1u && set_exec_lo == 1u &&
-              set_exec_hi == 1u && ballots == 0u && exec.IsImmediate() &&
-              exec.U1() && exec_mask_tag.IsImmediate() && exec_mask_tag.U1() &&
+    Check(set_exec == 1u && set_exec_lo == 1u && set_exec_hi == 1u &&
+              ballots == 0u && exec.IsImmediate() && exec.U1() &&
               exec_lo.IsImmediate() && exec_lo.U32() == 1u &&
               exec_hi.IsImmediate() && exec_hi.U32() == 0u,
           "typed entry is not the local {1,0} invocation mask");
@@ -9912,16 +9968,18 @@ void TestNewShaderRecompilerPerInvocationMasksWithoutMirrors() {
   result = ShaderRecompiler::Recompile(wqm_shader, options);
   CheckSpirvBinaryValidates(result.spirv);
   const auto wqm_source = DisassembleSpirvBinary(result.spirv);
-  Check(!Common::ContainsStr(wqm_source, "OpGroupNonUniformBallot"),
-        "raw scalar WQM reconstructed its source from invocation provenance");
+  Check(Common::ContainsStr(wqm_source, "OpCapability GroupNonUniformBallot") &&
+            Common::ContainsStr(wqm_source, "OpGroupNonUniformBallot"),
+        "per-invocation scalar WQM omitted its subgroup ballot capability");
+  Check(SpirvInstructionOpcodeCount(result.spirv, 132u) == 2u,
+        "wave64 WQM did not compact exactly two ballot words");
 
   auto wave32_options = options;
   wave32_options.wave_size = 32u;
   result = ShaderRecompiler::Recompile(wqm_shader, wave32_options);
   CheckSpirvBinaryValidates(result.spirv);
-  Check(!Common::ContainsStr(DisassembleSpirvBinary(result.spirv),
-                             "OpGroupNonUniformBallot"),
-        "wave32 raw scalar WQM reconstructed invocation provenance");
+  Check(SpirvInstructionOpcodeCount(result.spirv, 132u) == 1u,
+        "wave32 WQM retained the unused high ballot-word expansion");
 
   const uint32_t cross_lane_shader[] = {
       EncodeSop2(0x25, 126, 132, 128), // s_bfm_b64 exec, 4, 0
@@ -10234,7 +10292,6 @@ void TestRenderTargetReverseExportMapping() {
         "pixel shader cache identity omitted the render-target export mapping");
 
   regs.ps_regs.data_addr = reinterpret_cast<uint64_t>(shader);
-  regs.ps_regs.chksum = 0xf16ab6f000000001ull;
   ShaderMappedData mapped{};
   mapped.code_size_bytes = sizeof(shader);
   ShaderMapUserData(regs.ps_regs.data_addr, mapped);
@@ -10320,7 +10377,7 @@ void TestNewShaderRecompilerNativeBindingPlan() {
   SetImageTestFormat(&user_data, 6, Prospero::BufferFormat::k32UInt);
   auto options = MakeCompileOptions(ShaderType::Compute);
   options.dump_ir = true;
-  options.user_data = user_data.data();
+  options.user_data = user_data;
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   const auto *buffers = ShaderRecompiler::IR::FindBinding(
@@ -10703,7 +10760,7 @@ void TestTypedDescriptorRealCarryAndScalarLoads() {
 
   auto user_data = ImageTestUserData();
   auto options = MakeCompileOptions(ShaderType::Pixel);
-  options.user_data = user_data.data();
+  options.user_data = user_data;
   auto result = ShaderRecompiler::Recompile(inline_sampler_shader, options);
 }
 
@@ -11122,7 +11179,6 @@ void TestNewShaderRecompilerPixelPipelineEntry() {
 
   HW::PixelShaderInfo regs{};
   regs.ps_regs.data_addr = reinterpret_cast<uint64_t>(shader);
-  regs.ps_regs.chksum = 0x1234567800000001ull;
   ShaderMappedData mapped{};
   mapped.code_size_bytes = sizeof(shader);
   ShaderMapUserData(regs.ps_regs.data_addr, mapped);
@@ -11132,7 +11188,8 @@ void TestNewShaderRecompilerPixelPipelineEntry() {
   ShaderVertexInputInfo vertex_info{};
   ShaderPixelInputInfo input_info{};
   const auto params = PrepareProgram(regs, sh, vertex_info, mappings, input_info);
-  Check(params.hash == regs.ps_regs.chksum && params.code.data() == shader,
+  Check(params.hash == XXH3_64bits(params.code.data(), params.code.size_bytes()) &&
+            params.code.data() == shader,
         "pixel shader program parameters lost source identity");
 
   const uint32_t vcc_load_shader[] = {
@@ -11150,7 +11207,6 @@ void TestNewShaderRecompilerPixelPipelineEntry() {
 
   HW::PixelShaderInfo vcc_regs{};
   vcc_regs.ps_regs.data_addr = reinterpret_cast<uint64_t>(vcc_load_shader);
-  vcc_regs.ps_regs.chksum = 0x6306606500000001ull;
   vcc_regs.ps_regs.rsrc2.user_sgpr = 29;
   vcc_regs.ps_user_sgpr.value[27] = static_cast<uint32_t>(table_address);
   vcc_regs.ps_user_sgpr.value[28] = static_cast<uint32_t>(table_address >> 32u);
@@ -11189,9 +11245,7 @@ void TestComputeLdsAllocationIdentity() {
           "COMPUTE_PGM_RSRC2 LDS allocation units were not decoded");
     auto options = MakeCompileOptions(ShaderType::Compute);
     options.shader_hash = params.hash;
-    options.shader_base = params.Base();
-    options.user_data = params.user_data.data();
-    options.user_data_count = static_cast<uint32_t>(params.user_data.size());
+    options.user_data = params.user_data;
     options.input_info.compute = &input_info;
     options.wave_size = input_info.wave_size;
     options.scratch_dwords = input_info.scratch_size_dwords;
@@ -11257,10 +11311,7 @@ void TestComputeLdsAllocationIdentity() {
         "AGC per-thread scratch size was not propagated");
   auto scratch_options = MakeCompileOptions(ShaderType::Compute);
   scratch_options.shader_hash = scratch_params.hash;
-  scratch_options.shader_base = scratch_params.Base();
-  scratch_options.user_data = scratch_params.user_data.data();
-  scratch_options.user_data_count =
-      static_cast<uint32_t>(scratch_params.user_data.size());
+  scratch_options.user_data = scratch_params.user_data;
   scratch_options.input_info.compute = &scratch_info;
   scratch_options.wave_size = scratch_info.wave_size;
   scratch_options.scratch_dwords = scratch_info.scratch_size_dwords;
@@ -11432,11 +11483,9 @@ void TestPixelProgramCacheBindingIdentity() {
   Check(MakeStageStaticKey(no_depth_export) != MakeStageStaticKey(shifted_push),
         "pixel shader identity omitted graphics push-bank placement");
 
-  auto check_program_identity = [&](const uint32_t *shader,
-                                    uint64_t checksum) {
+  auto check_program_identity = [&](const uint32_t *shader) {
     HW::PixelShaderInfo regs{};
     regs.ps_regs.data_addr = reinterpret_cast<uint64_t>(shader);
-    regs.ps_regs.chksum = checksum;
     ShaderMappedData mapped{};
     mapped.code_size_bytes = sizeof(uint32_t);
     ShaderMapUserData(regs.ps_regs.data_addr, mapped);
@@ -11462,8 +11511,8 @@ void TestPixelProgramCacheBindingIdentity() {
     return std::pair {first_params.hash, first_key};
   };
 
-  const auto request_01 = check_program_identity(shader_01, 0);
-  const auto request_10 = check_program_identity(shader_10, 0);
+  const auto request_01 = check_program_identity(shader_01);
+  const auto request_10 = check_program_identity(shader_10);
   Check(request_01 == request_10,
         "relocated identical pixel programs did not share their source identity");
 
@@ -11474,7 +11523,6 @@ void TestPixelProgramCacheBindingIdentity() {
   };
   HW::PixelShaderInfo push_regs{};
   push_regs.ps_regs.data_addr = reinterpret_cast<uint64_t>(push_shader);
-  push_regs.ps_regs.chksum = 0x91a27e6300000003ull;
   push_regs.ps_regs.rsrc2.user_sgpr = 4;
   ShaderMappedData push_mapped{};
   push_mapped.code_size_bytes = sizeof(push_shader);
@@ -11537,8 +11585,7 @@ void TestGraphicsPushConstantPlacement() {
 
   auto options = MakeCompileOptions(ShaderType::Pixel);
   options.input_info.pixel = &pixel_info;
-  options.user_data = user_data.data();
-  options.user_data_count = static_cast<uint32_t>(user_data.size());
+  options.user_data = user_data;
   options.push_constant_offset = 36;
 
   auto placed = ShaderRecompiler::Recompile(shader, options);
@@ -11601,7 +11648,6 @@ void TestNewShaderRecompilerFlatUserPointerUsesDma() {
 
   auto options = MakeCompileOptions(ShaderType::Pixel);
   options.user_data = user_data;
-  options.user_data_count = static_cast<uint32_t>(std::size(user_data));
 
   auto result = ShaderRecompiler::Recompile(shader, options);
   Check(result.program.info.uses_dma,
@@ -11623,7 +11669,6 @@ void TestNewShaderRecompilerFlatAddressDomainsUseDma() {
 
   auto options = MakeCompileOptions(ShaderType::Compute);
   options.user_data = user_data;
-  options.user_data_count = static_cast<uint32_t>(std::size(user_data));
   options.scratch_dwords = 1;
   auto result = ShaderRecompiler::Recompile(segmented_shader, options);
   Check(result.program.info.uses_dma,
@@ -11856,8 +11901,8 @@ void TestNewShaderRecompilerSpirvSizeBaselines() {
                                    .conditional_branches = 1,
                                    .ballots = 1},
                                   ShaderType::Vertex);
-  Check(Common::ContainsStr(wqm_result.ir_dump, "WqmU64"),
-        "WQM size fixture no longer reaches raw WqmU64 IR");
+  Check(Common::ContainsStr(wqm_result.ir_dump, "WqmMask"),
+        "WQM size fixture no longer reaches per-invocation WqmMask IR");
 
   const uint32_t dispatcher[] = {
       EncodeSopp(0x05, 2),       // entry -> B, fallthrough A
@@ -11902,6 +11947,7 @@ int main() {
   TestNewShaderRecompilerSpirvSizeBaselines();
   TestDemandDrivenSpirvDeclarations();
   TestNewShaderRecompilerSMovB32();
+  TestNewShaderRecompilerClipDisabledPosition();
   TestNewShaderRecompilerAuxPositionExports();
   TestNewShaderRecompilerNativeWideScalarMemoryIr();
   TestNewShaderRecompilerNativeWideBufferIr();