AlxcBTW 350b8d2b88 fix(agc): keep the template of another address when a dispatch move is refused ...

Read-only addresses are not part of the dispatch key, so the fresh build after a refused
move replaced the template that serves the other address, and the two addresses
rebuilt each other. A refused dispatch now uses or builds an entry under the exact key
(read-only bases included) and leaves the movable template alone. The [dispmove] line
counts the rebuilds after a refusal, the builds inserted beside the template and the
dispatches served by an exact entry. Restores the MarkGpuWrites comment.

2026-10-17 13:52:16 +02:00

..

execution

fix(agc): keep the template of another address when a dispatch move is refused

2026-10-17 13:52:16 +02:00

shaders

Merge pull request #4408 from santhoshkumar-dev/codex/agc-render-target-vertex-aliasing

2026-10-13 16:29:52 +02:00

AsyncLog.cpp

fix(agc): flush the log queue, bounded, where the process ends without exit handlers

2026-10-15 20:42:16 +02:00

BdaContracts.cpp

fix(shader): record BDA faults in hull programs instead of refusing them

2026-10-14 04:59:27 +02:00

BdaDevice.cpp

test(recompiler): verify BDA store boundary faults and written pages

2026-10-11 07:46:41 -04:00

BdaExecution.cpp

fix(shader): store unaligned 32-bit BDA writes byte by byte

2026-10-12 16:06:16 +02:00

BdaResources.cpp

Merge pull request #5815 from EagleBearerYS/fix/agc-space-owner-overlap

2026-10-15 20:34:16 +02:00

BdaShader.cpp

fix(shader): store unaligned 32-bit BDA writes byte by byte

2026-10-12 16:06:16 +02:00

BdaShader.hpp

fix(shader): store unaligned 32-bit BDA writes byte by byte

2026-10-12 16:06:16 +02:00

BdaTests.hpp

test(agc): reproduce stale bytes in sparse buffer holes

2026-10-11 21:06:02 -04:00

BufferPool.cpp

feat(agc): place device buffers in host memory when device memory runs out

2026-10-13 22:46:59 +01:00

BufferResourceVcc.cpp

style(shader): trim descriptor fixture trailing blank lines

2026-10-14 12:26:01 +03:00

CmaskFastClear.cpp

fix(agc): keep the texels of CMASK tiles with codes other than 0 and 15

2026-10-13 02:17:33 +02:00

ColorTargetLayout.cpp

feat(agc): render into SW_64KB_Z_X color targets

2026-10-13 22:36:27 +01:00

ColorTransferTests.cpp

feat(agc): present 16-bit float display buffers

2026-10-11 00:38:12 +02:00

ColorTransferTests.hpp

…

ComputePipelineCapture.cpp

feat(agc): carry spanned waves in the target and require full subgroups

2026-10-16 14:12:58 +02:00

ComputePipelineCaptureRun.cmake

feat(agc): capture and replay compute pipeline compilation

2026-10-11 16:09:48 -04:00

ComputePipelineReplay.cpp

fix(agc): accept core Vulkan capabilities during compute replay

2026-10-11 16:34:26 -04:00

ConstantBitTest.cpp

fix(shader): fold only constant lower-half lane words of vertex programs

2026-10-15 07:38:42 -04:00

ConstantPhi.cpp

fix(shader): preserve distinct definitions while folding phi inputs

2026-10-11 16:44:34 -03:00

ControlFlow.cpp

Merge pull request #4840 from Lorkydey/feat/shader-fragment-user-data-tail

2026-10-14 14:56:34 +02:00

DccKeySharing.cpp

…

DenormalFlushElimination.cpp

perf(shader): treat f32 min, max, med3, floor and ceil results as denormal-free

2026-10-12 09:24:29 -04:00

DepthSurfaceReuse.cpp

fix(agc): treat a view past a depth surface's first level as memory

2026-10-15 16:55:31 +00:00

DeviceFault.cpp

feat(agc): report device losses with VK_EXT_device_fault and dispatch checkpoints

2026-10-14 13:50:52 +02:00

DeviceMemoryFallback.cpp

fix(agc): select coherent buffer heaps before allocation

2026-10-14 15:48:04 +13:00

DisplayBufferDcc.cpp

feat(agc): present BT.2100 PQ display buffers on an SDR output

2026-10-14 02:55:21 -04:00

DppWrite.cpp

…

DrawProgramTrace.cpp

feat(agc): trace shader addresses for each draw

2026-10-14 23:39:27 +02:00

EmbeddedVertexFetch.cpp

fix(recompiler): apply the S_BFE_U32 width when the vertex fetch scanner folds constants

2026-10-09 10:37:40 -06:00

EmbeddedVertexFetchAliasing.cpp

fix(shader): preserve aliased embedded fetch attributes

2026-10-10 07:22:51 -03:00

ExecutionTestNotBuilt.cpp

ci(windows): build only the execution tests that run without a Vulkan device

2026-10-13 13:44:08 +02:00

Flip.cpp

fix(agc): print the stored worker failure before WaitIdle can return

2026-10-14 19:24:49 +00:00

Float16Conversion.cpp

perf(shader): map cached SPIR-V from the shader cache files

2026-10-14 10:36:24 +04:00

FmaMixOnce.cpp

refactor(recompiler): emit the v_fma_mixlo/hi and v_interp_*_f16 emulation once per module

2026-10-14 11:20:43 -04:00

FormattedBufferByteStores.cpp

perf(shader): store 8- and 16-bit buffer components as bytes in compute

2026-10-13 16:36:56 +02:00

FormattedBufferOnce.cpp

fix(shader): decode a GPU-selected formatted load from a format table

2026-10-15 10:32:47 +13:00

FragmentBarycentric.cpp

…

FrameDumpIo.cpp

test(agc): write more pixels than a pipe holds in the closed-pipe case

2026-10-15 14:04:10 -07:00

Graphics.cpp

fix(agc): compare moved dispatch buffers at their build-time clamped size

2026-10-17 11:23:47 +02:00

GraphicsTests.hpp

fix(agc): keep detiler descriptor pools unrecycled without a recorder

2026-10-15 10:14:29 +02:00

GuestAllocations.cpp

perf(libc): share one snapshot of the registered guest ranges between leases until the registry changes

2026-10-12 21:07:13 +02:00

GuestSamplerResource.cpp

fix(agc): accept a min or max reduction sampler with anisotropic filtering

2026-10-15 15:16:41 +02:00

GuestTextureResource.cpp

fix(agc): treat a TYPE 1 T# in a read-only static image slot as null

2026-10-15 15:16:56 +01:00

HostHdr.cpp

feat(videoout): present PQ display buffers through an HDR10 swapchain

2026-10-14 02:55:21 -04:00

HostImportRetireHold.cpp

fix(agc): promote a held import window to its whole allocation

2026-10-14 15:19:28 +01:00

HostImportSparseDmaBuf.cpp

Merge pull request #6261 from EdouardSence/fix/sparse-import-test-budget

2026-10-17 16:48:16 +13:00

IntegerNarrowing.cpp

fix(shader): preserve integer narrowing in constant folding

2026-10-12 06:08:56 +05:30

IrReclaim.cpp

perf(shader): keep only evaluable branch conditions in resource plans

2026-10-15 08:22:09 +13:00

KeyMiss.cpp

feat(agc): implement image and sampler tables indexed at runtime offsets

2026-10-14 16:43:57 +02:00

LiveStack.cpp

fix(agc): report guest addresses above the user address space as unmapped on Windows

2026-10-13 17:56:49 -03:00

MaskedSelect.cpp

refactor(agc): separate stencil replay from its follow-up topics

2026-10-14 22:08:48 +02:00

Mesh.cpp

fix(agc): validate mesh index descriptors on cached draws (reapply #5791)

2026-10-15 22:43:15 +02:00

MeshGeometryFixture.hpp

fix(agc): validate mesh index descriptors on cached draws (reapply #5791)

2026-10-15 22:43:15 +02:00

MubufLdsDecode.cpp

…

NarrowConstantStoreFixture.hpp

…

PassHazards.cpp

feat(agc): render pass hazard tracker

2026-10-10 23:27:18 +13:00

PerformanceTrace.cpp

feat(agc): report which trace threads dropped events and where their buffers filled

2026-10-14 18:54:44 +02:00

PipelineLinkOptimization.cpp

fix(agc): skip the background pipeline relink on NVIDIA's proprietary driver

2026-10-11 12:44:46 +02:00

PixelPositionExport.cpp

…

PlaceholderImages.cpp

fix(agc): typed image heaps on devices without nullDescriptor

2026-10-11 10:32:49 +02:00

Pm4.cpp

feat(agc): keep the last submissions per queue and dump them on a stall or crash

2026-10-16 11:18:26 +02:00

Pm4Replay.cpp

feat(agc): capture submissions and replay their decode and wait releases offline

2026-10-13 22:40:11 -04:00

Pm4ReplayTool.cmake

feat(agc): capture submissions and replay their decode and wait releases offline

2026-10-13 22:40:11 -04:00

PositionInvariance.cpp

fix(shader): clip primitives with an all-zero clip position

2026-10-12 06:46:34 +02:00

PresentationPass.cpp

feat(videoout): present PQ display buffers through an HDR10 swapchain

2026-10-14 02:55:21 -04:00

ProfileOutput.cpp

…

RdnaTruncatedModifier.cpp

…

Recorder.cpp

Merge pull request #6225 from Amoenus/codex/test-recorder-gpu-selection

2026-10-17 16:33:40 +13:00

RecorderKeptBytes.cpp

fix(agc): discard recorder batches after submission failure

2026-10-15 06:06:00 +00:00

RectListPositionInvariant.cpp

fix(shader): decorate rect-list gl_Position as Invariant

2026-10-12 10:45:38 +02:00

RedundantValueElimination.cpp

feat(recompiler): reuse pure IR values recomputed in dominated blocks

2026-10-13 22:40:56 -04:00

RegisterKey.cpp

fix(agc): mix shader identity into trace key

2026-10-15 13:06:28 +02:00

ResidentPresent.cpp

perf(agc): present 8-bit integer resident images without a readback

2026-10-14 22:04:55 +03:00

ResidentPresent.hpp

…

ScalarSelectMask.cpp

fix(shader): keep s_cselect_b64 of two masks as a lane bit and fold constant masks

2026-10-10 10:15:05 +02:00

ShaderDeviceProfile.cpp

fix(recompiler): hold 32 samplers in the runtime sampler heap

2026-10-17 16:36:17 +13:00

ShaderDiskCache.cpp

feat(agc): carry spanned waves in the target and require full subgroups

2026-10-16 14:12:58 +02:00

ShaderMemory.cpp

feat(agc): carry spanned waves in the target and require full subgroups

2026-10-16 14:12:58 +02:00

ShaderModeRegisters.cpp

feat(shader): track MODE rounding changes through the control flow graph

2026-10-14 15:07:24 +02:00

ShaderReplayRequests.cpp

feat(agc): replay prepared shader capture and materialization offline

2026-10-16 20:34:23 +02:00

ShaderReplayTool.cmake

feat(agc): replay prepared shader capture and materialization offline

2026-10-16 20:34:23 +02:00

SharedWrites.cpp

fix(libc): track writes to shared guest memory per 4 KiB page on Windows

2026-10-12 18:03:53 -04:00

Spirv13Image.cpp

…

SpirvBranchTargets.cpp

perf(shader): map cached SPIR-V from the shader cache files

2026-10-14 10:36:24 +04:00

SpirvLdsAtomicLock.cpp

fix(shader): 32-bit flat atomics in the LDS aperture take the workgroup lock

2026-10-11 01:53:37 +03:00

SpirvModule.cpp

perf(shader): index specialization tables by SPIR-V id

2026-10-13 07:59:08 +01:00

SpirvNarrowConstantStore.cpp

test(agc): validate the narrow constant store fixture through its word pointer

2026-10-14 21:56:47 +13:00

SpirvOptimizer.cpp

Merge pull request #5511 from JJenkx/perf/spirv-bounded-redundancy

2026-10-15 07:38:13 +13:00

StackWalk.cpp

feat(agc): write the worker sampler's stacks as collapsed stacks

2026-10-12 21:23:11 +02:00

Structurizer.cpp

test(recompiler): cover slow shared merge splitting and its growth budget

2026-10-14 01:00:21 +02:00

SubgroupClock.cpp

fix(agc): detect narrow clock on RX 7800 XT

2026-10-09 22:04:49 +00:00

Submit.cpp

fix(agc): read unregistered graphics programs beside registered shaders

2026-10-15 06:24:14 +05:30

TessellationFactorRing.cpp

…

TextureDetiler.cpp

fix(agc): keep detiler descriptor pools unrecycled without a recorder

2026-10-15 10:14:29 +02:00

TextureFormat.cpp

Merge pull request #5149 from xXPBSXx/fix/agc-texture-snorm-sint-2101010

2026-10-14 14:34:11 +13:00

TextureTiling.cpp

Merge pull request #1639 from gerardlesan/fix/agc-zx-mip-tail-limits

2026-10-10 17:40:18 +13:00

ThreadPriority.cpp

fix(agc): report a refused queue worker priority as a driver failure and reject unknown APS5_THREAD_PRIORITY values

2026-10-10 12:23:12 -06:00

ValidateProgram.cpp

…

VectorWindow.cpp

test(shader): make the read-only window tests pass in the full Linux build

2026-10-13 14:09:05 +13:00

VertexBarrier.cpp

fix(shader): clip primitives with an all-zero clip position

2026-10-12 06:46:34 +02:00

VertexInput.cpp

fix(agc): copy an indirect draw's vertex buffer only up to its last element

2026-10-15 02:54:23 +00:00

VertexLayerExport.cpp

perf(shader): map cached SPIR-V from the shader cache files

2026-10-14 10:36:24 +04:00

Visual.cpp

fix(agc): restore the standalone visual demo

2026-10-09 14:12:24 +02:00

Vop1ControlDecode.cpp

…

Vop1SdwaMoveDecode.cpp

…

Vop2EncodingDecode.cpp

fix(shader): stop decoding gfx6/7-only VOP2 opcodes on RDNA

2026-10-09 12:30:40 -04:00

Vop3InterpDecode.cpp

fix(shader): decode the high-half result of v_interp_p2_f16

2026-10-13 20:37:44 +00:00

WaitShutdown.cpp

fix(agc): wait for shutdown at shader ABI entrypoints

2026-10-11 12:48:40 +00:00

WaveControlDecode.cpp

feat(shader): add GDS ordered counts and ordered-append dispatches

2026-10-14 21:23:52 +02:00

WriteTracking.cpp

feat(agc): prove complete tracked driver store coverage

2026-10-15 14:00:57 +02:00

WriteWatchCoverage.cpp

fix(agc): save pending storage results in every 64 KiB block that leaves the write watch

2026-10-15 14:01:01 +02:00