Commit Graph

609 Commits

Author SHA1 Message Date
Matthew Hurley 5a705dd15f shader: support V_CMPX_EQ_U16 (#816)
* shader: support V_CMPX_EQ_U16

Decode gfx11 VOPC 0xba and lower it as an unsigned 16-bit equality
compare with CMPX EXEC-mask semantics.

🤖 Generated with Codebuff 🤖
Co-Authored-By: Codebuff <noreply@codebuff.com>

* shader: correct CMPX EQ U16 runtime expectations

Remove the added CFG tests and correct two EXEC expectations in the compute regression case.

---------

Co-authored-by: Matthew Hurley <171001439+theantipopau@users.noreply.github.com>
Co-authored-by: Codebuff <noreply@codebuff.com>
Co-authored-by: nmzik <Nmzik@mail.ru>
2026-09-25 02:27:44 +04:00
Saadi b9a0bd2791 loader: wrap elapsed time in Timer::GetTime (#679)
loader: format elapsed log timestamps beyond 24 hours

Co-authored-by: nmzik <Nmzik@mail.ru>
2026-09-24 22:02:19 +04:00
nmzik d72a79af2d graphics: skip compute dispatches with BVH instructions
Co-authored-by: Mac <me@itsmemac.com>
2026-09-24 20:46:04 +04:00
nmzik dc9280d75d shader: add GLC/DLC buffer cache policy support
Co-authored-by: Mac <me@itsmemac.com>
2026-09-24 20:46:04 +04:00
nmzik 6a9ae4ae43 shader: add SDWA support for V_MAX_I32 2026-09-24 20:46:04 +04:00
nmzik 392f39e2f6 shader: support indirect BUFFER_LOAD_DWORDX3 2026-09-24 20:46:04 +04:00
nmzik e43c242683 shader: add S_MEMREALTIME opcode
Co-authored-by: Maxime Allanic <maxime@allanic.me>
Co-authored-by: Mac <me@itsmemac.com>
2026-09-24 20:46:04 +04:00
Douglas Soares Silva fcf0f3b67e shader: fold the lane test after a pixel shader's entry s_wqm (#801)
Preserve live lane predicates through RDNA2 whole-quad expansion

Keep the source-live predicate when WQM expands a scalar mask, so entry masks remain known live without an instruction pre-scan. Preserve scalar mask values, SCC, half writes, and raw-source behavior.

Cover entry pixel masks, cross-half copies, and scalar/high-word sources with existing shader regression suites.

Co-authored-by: nmzik <Nmzik@mail.ru>
Co-authored-by: Douglas <douglasproducoes21@gmail.com>
2026-09-24 12:02:19 +04:00
nmzik 84d015120c Remove duplicate image scratch setup from push constant test 2026-09-24 11:13:33 +04:00
nmzik a6248c7c7c Convert formatted half-float buffer stores from F32
Co-authored-by: TarkusR <90155059+TarkusR@users.noreply.github.com>
2026-09-24 11:13:33 +04:00
nmzik 3bd5de646d Reuse prepared shader bindings across draws and dispatches 2026-09-24 11:13:33 +04:00
nmzik 0382d0ca58 Publish reserved guest mappings atomically after backing commit 2026-09-24 11:13:33 +04:00
nmzik a670dc84b3 Test RDNA2 wave32 ReadLane material key selection 2026-09-24 11:13:33 +04:00
nmzik 963a0817dc Bound uniformized image keys by guarded material mask
Co-authored-by: TarkusR <90155059+TarkusR@users.noreply.github.com>
2026-09-24 11:13:33 +04:00
nmzik 66cc2fe26d Track depth and stencil attachment feedback across graphics stages
Co-authored-by: TarkusR <90155059+TarkusR@users.noreply.github.com>
2026-09-24 11:13:33 +04:00
nmzik 51c34d1a24 Test zero scalar shift folding on RDNA2 compute shader 2026-09-24 11:13:33 +04:00
nmzik 3b89e36d1e Track bounded indirect image tables from signed shader loops
Co-authored-by: TarkusR <90155059+TarkusR@users.noreply.github.com>
2026-09-24 11:13:33 +04:00
João Lobo 3dc3fd0197 shader: implement DS_INC_U32 and DS_DEC_U32 (#790) 2026-09-24 00:45:12 +04:00
Deniz ŞAHİN fc86509aac shader: support DS_WRITE_B8_D16_HI (#794)
Decode GFX10 DS opcode 0xa0 as DS_WRITE_B8_D16_HI (next to
DS_WRITE_B16_D16_HI at 0xa1) and lower it through the existing 8-bit
shared-memory write. The stored byte is bits 23:16 of the data VGPR,
selected with SDWA BYTE_2 on src1.

Add captured-instruction decode coverage (from GTA V, PPSA04264) and a
runtime regression that writes one byte of a distinctive source value
into a pre-filled LDS dword.
2026-09-24 00:44:39 +04:00
nmzik 569ec04d1a Fix save metadata and timestamps
Co-authored-by: Ali Almohaya <AliAlmohaya@gmail.com>
Co-authored-by: CheesyPoofs346 <jjs777406@gmail.com>
2026-09-23 22:50:51 +04:00
nmzik 7d7267cc56 Translate NGG mesh prolog with guest wave32 width 2026-09-23 22:50:51 +04:00
nmzik f68c77d785 SDL3 update (#787)
* Update to SDL3

* Add XTest development dependency for SDL3

* Preserve native entry points with SDL3

* Document SDL3 XTest dependency
2026-09-23 17:22:15 +04:00
nmzik 9860c2bbf8 Associate stencil when depth is acquired
Co-authored-by: Daniil Shushakov <4shushakov@gmail.com>
2026-09-23 15:41:02 +04:00
nmzik ee3749d9b6 Cover 64 MiB download ring staging 2026-09-23 15:41:02 +04:00
nmzik 79530b4c5e Read indirect dispatch counts from GPU buffers
Co-authored-by: Mac <me@itsmemac.com>
2026-09-23 15:41:02 +04:00
nmzik d8a4412498 Refresh stencil backing through the depth image
Co-authored-by: Daniil Shushakov <4shushakov@gmail.com>
2026-09-23 15:41:02 +04:00
nmzik 5ed5fdd36e Preserve resource level bits in texture descriptors 2026-09-23 15:41:02 +04:00
nmzik ef83ba65ad Validate texture descriptor fields 2026-09-23 15:41:02 +04:00
nmzik d30dca78be Read image descriptors in one bounded span 2026-09-23 15:41:02 +04:00
nmzik 66f1a82365 Resolve scalar image tables during resource planning 2026-09-23 15:41:02 +04:00
nmzik eb8a5504a3 Emit coordinates for each indirect image candidate
Co-authored-by: Techx3 <56268269+Techx3@users.noreply.github.com>
2026-09-23 15:41:02 +04:00
nmzik 03c55e5712 Keep shader user data inline for cached programs 2026-09-23 15:41:02 +04:00
nmzik 5d1ad31c49 Resolve guarded descriptor Phis at their use edge
Co-authored-by: Techx3 <56268269+Techx3@users.noreply.github.com>
Co-authored-by: GodsWill <GodsWill@users.noreply.github.com>
2026-09-23 15:41:02 +04:00
nmzik 13fcb3a133 Refresh cached resources in one SRT walk 2026-09-23 15:41:02 +04:00
nmzik d6dac92da1 Reuse scalar evaluation storage
Co-authored-by: MehmetCambaz <mcambaz18@gmail.com>
Co-authored-by: Leclowndu93150 <xdr.a123454321@gmail.com>
2026-09-23 15:41:02 +04:00
nmzik 90471836f7 Resolve storage mip views before layout lookup 2026-09-23 15:34:44 +04:00
nmzik 8dc9b9d532 Persist save allocations for search and mount capacity queries 2026-09-22 11:38:50 +04:00
nmzik 7c9e18e73b Ignore inactive metadata address bits in sampled depth descriptors
Co-authored-by: Ali Almohaya <17254073+Almo7aya@users.noreply.github.com>
2026-09-22 11:38:50 +04:00
nmzik d2df2cad74 Resolve draw-uniform image descriptors through scalar buffer chains 2026-09-22 11:38:50 +04:00
nmzik b2a80a3a96 Resolve GPU-selected raw buffer descriptors through shared RDNA2 addressing
Co-authored-by: Bipin <87369440+bipinkrish@users.noreply.github.com>
2026-09-22 11:38:50 +04:00
nmzik cfb8f7b246 Decode unsigned byte sources for SDWA half-float conversion 2026-09-22 11:11:36 +04:00
nmzik 476204cd4b Initialize native NGG vertex launch state and wave size 2026-09-22 11:11:36 +04:00
nmzik f9e56ef8a0 Scope native barriers to shaders with shared execution 2026-09-22 11:11:36 +04:00
nmzik ba55ba5dc5 Allow storage access to PS5 cubemap faces 2026-09-21 00:49:54 +04:00
Elija 549495575c shader: add S_BITCMP0_B64 and S_BITCMP1_B64 (#688)
* shader: add S_BITCMP0_B64 and S_BITCMP1_B64

* test: move S_BITCMP_B64 coverage to compute suite

* test: use the existing compute CTest entry for bit comparisons

* test: remove the dedicated S_BITCMP_B64 command-line selector

---------

Co-authored-by: nmzik <Nmzik@mail.ru>
2026-09-20 19:36:12 +04:00
Maayan080 c3cc8454b9 common: remove unused recursive file enumeration (#681)
common: don't follow reparse points in the Windows recursive file walk

SysFileFindFiles recurses into every entry carrying
FILE_ATTRIBUTE_DIRECTORY, which on Windows includes directory symlinks
and junctions. A junction pointing at one of its own ancestors makes the
walk recurse until the stack runs out.

The Linux implementation already guards against this: it uses lstat
rather than stat specifically so a symlink is never followed into a
cycle. Skip reparse-point entries on Windows for the same reason.

Co-authored-by: nmzik <Nmzik@mail.ru>
2026-09-20 18:50:57 +04:00
Elija cfb110e0aa shader: allow partial SDWA destinations for V_NOT_B32 (#684) 2026-09-20 18:21:40 +04:00
nmzik 0ad9149c90 Fix three-vertex rectangle rendering 2026-09-20 17:16:19 +04:00
nmzik 102a10f0fe Fix signed sampler comparisons 2026-09-20 17:16:19 +04:00
nmzik 8cecaf4e30 Merge PR #716: Fix Unicode paths on Windows
Preserve native Windows paths for filesystem operations, command-line options,
and game icons. Complete Unicode handling in recursive file listings and file
logging, restore unmounting by host directory, and extend regression coverage.

Original implementation by Augusto Domingos (@AghastyGD); retain all five of
his PR commits and include the reviewed follow-up fixes.

Closes #714

Co-authored-by: Augusto Domingos <augustodomingosvasco@gmail.com>
2026-09-20 02:14:22 +04:00