shader: implement RDNA2 DS forward lane permute

Co-authored-by: shadowbeat070 <70843316+shadowbeat070@users.noreply.github.com>
This commit is contained in:
nmzik
2026-10-04 09:26:44 +04:00
parent 46900d8042
commit c17fa5b936
10 changed files with 147 additions and 6 deletions
@@ -116,7 +116,7 @@ constexpr MemoryOpcodeInfo DS_OPCODE_LIST[] = {
{0xa6u, Opcode::DS_READ_U16_D16, 1, 16},
{0xa7u, Opcode::DS_READ_U16_D16_HI, 1, 16},
{0xb0u, Opcode::DS_WRITE_ADDTID_B32, 1, 32}, {0xb1u, Opcode::DS_READ_ADDTID_B32, 1, 32},
{0xb3u, Opcode::DS_BPERMUTE_B32, 1, 32},
{0xb2u, Opcode::DS_PERMUTE_B32, 1, 32}, {0xb3u, Opcode::DS_BPERMUTE_B32, 1, 32},
{0xdeu, Opcode::DS_WRITE_B96, 3, 32}, {0xdfu, Opcode::DS_WRITE_B128, 4, 32},
{0xfeu, Opcode::DS_READ_B96, 3, 32}, {0xffu, Opcode::DS_READ_B128, 4, 32},
};
@@ -205,6 +205,7 @@ uint32_t DsSourceCount(Opcode opcode) {
case Opcode::DS_WRITE2ST64_B64:
case Opcode::DS_MIN_F32:
case Opcode::DS_MAX_F32: return 3u;
case Opcode::DS_PERMUTE_B32:
case Opcode::DS_BPERMUTE_B32: return 2u;
case Opcode::DS_READ_ADDTID_B32:
case Opcode::DS_CONSUME:
@@ -407,7 +408,8 @@ void DecodeDs(uint32_t pc, std::span<const uint32_t> code, uint32_t word_index,
SetUnsupported(inst, Family::DS, opcode, "DS swizzle FFT mode is not implemented");
}
if (inst.gds &&
(inst.opcode == Opcode::DS_SWIZZLE_B32 || inst.opcode == Opcode::DS_BPERMUTE_B32 ||
(inst.opcode == Opcode::DS_SWIZZLE_B32 || inst.opcode == Opcode::DS_PERMUTE_B32 ||
inst.opcode == Opcode::DS_BPERMUTE_B32 ||
inst.opcode == Opcode::DS_WRITE_ADDTID_B32 ||
inst.opcode == Opcode::DS_READ_ADDTID_B32)) {
SetUnsupported(inst, Family::DS, opcode, "DS lane operation is available only for LDS");
@@ -701,6 +701,7 @@ std::string InstructionToString(const Instruction& inst) {
case Opcode::DS_MIN_F32:
case Opcode::DS_MAX_F32:
case Opcode::DS_SWIZZLE_B32:
case Opcode::DS_PERMUTE_B32:
case Opcode::DS_BPERMUTE_B32:
case Opcode::DS_READ_I8:
case Opcode::DS_READ_U8:
@@ -562,6 +562,7 @@ enum class Opcode {
DS_MIN_F32,
DS_MAX_F32,
DS_SWIZZLE_B32,
DS_PERMUTE_B32,
DS_BPERMUTE_B32,
DS_CONSUME,
DS_APPEND,
@@ -894,9 +894,10 @@ void Translator::DS_SWIZZLE_B32(const Decoder::Instruction& inst) {
ReadU32(MemorySourceAt(inst, 1)), ir.GetExec()}));
}
void Translator::DS_BPERMUTE_B32(const Decoder::Instruction& inst) {
void Translator::DS_PERMUTE(const Decoder::Instruction& inst, bool backward) {
const auto address = ir.IAdd(ReadU32(inst.src0), IR::U32(IR::Value(inst.offset)));
WriteOperand(inst.dst, ir.Emit(IR::ValueOpcode::BpermuteU32,
WriteOperand(inst.dst, ir.Emit(backward ? IR::ValueOpcode::BpermuteU32
: IR::ValueOpcode::PermuteU32,
{ReadU32(inst.src1), address, ir.GetExec()}));
}
@@ -1112,7 +1113,8 @@ void Translator::EmitMemory(const Decoder::Instruction& inst) {
case Decoder::Opcode::DS_MAX_F32:
return DS_MINMAX_F32(inst, IR::ValueOpcode::SharedAtomicFMax32);
case Decoder::Opcode::DS_SWIZZLE_B32: return DS_SWIZZLE_B32(inst);
case Decoder::Opcode::DS_BPERMUTE_B32: return DS_BPERMUTE_B32(inst);
case Decoder::Opcode::DS_PERMUTE_B32: return DS_PERMUTE(inst, false);
case Decoder::Opcode::DS_BPERMUTE_B32: return DS_PERMUTE(inst, true);
case Decoder::Opcode::DS_CONSUME:
return DS_APPEND_CONSUME(inst, IR::ValueOpcode::DataConsume);
case Decoder::Opcode::DS_APPEND:
@@ -109,7 +109,7 @@ private:
void DS_APPEND_CONSUME(const Decoder::Instruction& inst, IR::ValueOpcode opcode);
void DS_ADDTID(const Decoder::Instruction& inst, bool write);
void DS_SWIZZLE_B32(const Decoder::Instruction& inst);
void DS_BPERMUTE_B32(const Decoder::Instruction& inst);
void DS_PERMUTE(const Decoder::Instruction& inst, bool backward);
IR::F32 SelectF32(IR::U1 condition, IR::F32 true_value, IR::F32 false_value);
IR::U32 ConvertF32ToU32Saturated(IR::F32 value, float upper_bound, float safe_upper,