fix(shader): recognize scalarized constant-stride image tables by Prophet069 · Pull Request #13149 · boykopovar/AnyPS5

Prophet069 · 2026-10-30T04:16:03Z

What

A PPSA02015 compute shader scales an image-table offset by a constant stride and then moves it through v_readfirstlane inside selects guarded by the lane id. Resource analysis rejected it with GetImageResource dword 0 is not a valid runtime value. The resource tracker now recognizes such an offset as a constant-stride column through selects whose condition the active mask implies. Runtime SRT evaluation gains unsigned greater-than-or-equal comparisons, uniform selected components of constructed U32x4 values and known active-mask conjuncts in nested selects. Image dimension queries stay in GPU IR, and variable scales are still not constant-stride columns. The new entry in docs/dev/debt/shader.md records that the live GPU behaviour of these additions is unmeasured.

Tested

With ANYPS5_REQUIRE_VULKAN=1, ANYPS5_NO_SHADER_CACHE=1 and VK_INSTANCE_LAYERS=VK_LAYER_KHRONOS_validation (VVL vulkan-sdk-1.4.321.0; neither log contains was not found but was requested by env var VK_INSTANCE_LAYERS), ctest --test-dir build -R <the 63 tests of agc_shader_memory_tests, agc_driver_cmask_fast_clear_tests, agc_driver_graphics_tests and agc_driver_sample_multisample_target_tests> -V on main 8c8b998620 and on this branch, same machine and GPU: main 63/63 passed, 0 VUID-; this branch 63/63 passed, 0 VUID-. New ShaderMemory.cpp cases cover unrelated masks, disjunctions, varying and out-of-range components and variable shifts.
agc_driver_unclamped_normalized_blend (the feat(agc): merge a draw into more than one unclamped normalized color target #12916 multi-target regression, 2 draws blending two targets each) passed on main and on this branch with the same validation settings, 0 VUID-.
agc_shader_replay --prepare of a captured PPSA02015 request (code hash 3d6b9e898ab900ba, 280 code words, wave64) with ANYPS5_NO_SHADER_CACHE=1: on main 8c8b998620 it exits 1 with resource analysis failed: GetImageResource dword 0 is not a valid runtime value; chain: LoadAddressU32@pc=0x00000208(…ReadFirstLane(SelectU32(ULessThan32(GetBuiltin…; on this branch it exits 0 with prepared: 133909 SPIR-V words. The request contains game code and is not attached.
The commit changes only the files listed in the diff; it no longer carries the pre-feat(agc): merge a draw into more than one unclamped normalized color target #12916/fix(agc): don't release a label wait on a store its queue already took and reset #12923 tree of the first push.

Checklist

🤖 Generated with Claude Code

MarcZX10

Reviewed — changes requested

The shader repro is author-reported as fixed, but the PR also removes multi-target color merging that is present on its base.

What I ran

Reviewed head 6bf492e804cefd0ef70ab4f4a062aeaa5eeed02d; PR base 5ecef095b61e7a406b279c97896f3a8a6fbd6353. Current main is 422f96b3a92e797c5f69c136c4a394c8da672556; the merge preview is clean (tree 300d4e4445d31e7887adfe6179550a5467fc4b87). Environment: Linux 6.1.0-54-amd64 x86_64. I did not build or run tests locally.

git diff --check upstream/main...upstream/pr-13149
git merge-tree --write-tree upstream/main upstream/pr-13149
python3 tools/check_conventions.py --base upstream/main --head upstream/pr-13149
git diff --unified=4 upstream/main...upstream/pr-13149 -- core/libs/prx/libSceAgcDriver/Graphics/src/State.cpp core/shader/recompiler/Recompiler.hpp core/shader/recompiler/Recompiler.cpp
python3 - <<'PY'
import subprocess
base='5ecef095b61e7a406b279c97896f3a8a6fbd6353'
main='422f96b3a92e797c5f69c136c4a394c8da672556'
head='6bf492e804cefd0ef70ab4f4a062aeaa5eeed02d'
def paths(a,b): return set(subprocess.check_output(['git','diff','--name-only',a,b],text=True).splitlines())
print(sorted(paths(base,main) & paths(base,head)))
PY
gh pr checks 13149 -R boykopovar/AnyPS5

Results

Check	Result
Diff check	Passed; no output
Current-main merge preview	Clean
Conventions	Passed locally and in hosted CI
Hosted Windows build, Ubuntu and Windows relinker, relinker ASan	Passed
Hosted macOS relinker jobs	Pending
Hosted Linux and register-context/write-watch jobs	Skipped
Author-reported GPU suite	63/63 on main and 63/63 on the change, same Windows 10 / RTX 3080 system, with no VUID output; not independently run
Author-reported shader repro	On main `08a10d2253`, preparation exits 1 with the reported invalid-runtime-value error; on the PR branch it exits 0 with 133,909 SPIR-V words; not independently run
Changes on current main since PR base	Four paths; none overlap the PR’s changed paths

Captured command output

300d4e4445d31e7887adfe6179550a5467fc4b87
All conventions checks passed.
[]
linux	skipping	0	https://github.com/boykopovar/AnyPS5/actions/runs/41001942789/job/123888190767
register-context	skipping	0	https://github.com/boykopovar/AnyPS5/actions/runs/41001942789/job/123888191171
write-watch	skipping	0	https://github.com/boykopovar/AnyPS5/actions/runs/41001942789/job/123888190820
intake	pass	17s	https://github.com/boykopovar/AnyPS5/actions/runs/41001942668/job/123888122972
relinker (ubuntu-latest, g++)	pass	3m16s	https://github.com/boykopovar/AnyPS5/actions/runs/41001942789/job/123888189359
relinker (windows-latest, g++)	pass	8m59s	https://github.com/boykopovar/AnyPS5/actions/runs/41001942789/job/123888189343
relinker-asan	pass	2m46s	https://github.com/boykopovar/AnyPS5/actions/runs/41001942789/job/123888189247
windows	pass	18m51s	https://github.com/boykopovar/AnyPS5/actions/runs/41001942789/job/123888189236
relinker (macos-latest, clang++)	pending	0	https://github.com/boykopovar/AnyPS5/actions/runs/41001942789/job/123888189285
relinker (macos-latest, clang++, x86_64)	pending	0	https://github.com/boykopovar/AnyPS5/actions/runs/41001942789/job/123888189379
conventions	pass	13s	https://github.com/boykopovar/AnyPS5/actions/runs/41001942682/job/123888122893
eligibility	pass	4s	https://github.com/boykopovar/AnyPS5/actions/runs/41001942789/job/123888122754
labels	pass	9s	https://github.com/boykopovar/AnyPS5/actions/runs/41001942778/job/123888123435
report	pass	41s	https://github.com/boykopovar/AnyPS5/actions/runs/41001942830/job/123888122166

Negative control

Assertion	Control and change	Outcome
Captured PPSA02015 shader preparation	Author reports baseline `08a10d2253` versus the PR branch	Baseline failed with the cited runtime-value error; PR branch prepared successfully. This is author-provided, not an independently run negative control.
Multi-target regression	No local test mutation was run	Not exercised locally; the source change itself rejects a second merged target.

Findings

Blocking

Restores a one-target limit and removes merged-target handling. Current main supports multiple partial-write/unclamped-normalized target merges: ExportTargetMerges returns a vector, and the recompiler materializes and hashes all entries. This PR changes that to a single TargetMerge and adds Require(merge.mask == 0, ...), which rejects a draw with a second mergeable target. The PR description does not identify this behavior change, and the change removes support introduced by #12916. Please preserve the multi-target state, materialization, cache and SPIR-V paths and their tests while applying the scalarized image-table fix.

Non-blocking

None.

Sources

Not verified

I did not run the author's Windows/Vulkan tests or shader replay locally, a local C++ build, Linux product tests, macOS relinker checks, or console behavior. The GPU and replay results above are the author's report.

To get this to a good state

Preserve multi-target merge handling and coverage from current main, then rerun the relevant multi-target regressions and the captured shader preparation comparison on the revised head. I will repeat the same source, diff, conventions and merge-preview checks on that head.

github-actions · 2026-10-30T04:47:15Z

Some comparisons could not be completed because a pull request conflicts with the base branch. Review these shared files manually before concluding that there is no overlap.

Pull request	Shared files not compared
PR 5021	`docs/dev/debt/shader.md`
PR 5371	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `docs/dev/debt/shader.md`
PR 5609	`core/shader/recompiler/Optimization/src/SrtWalker/SrtEvaluator.cpp`
PR 6235	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `docs/dev/debt/shader.md`
PR 6280	`core/shader/recompiler/Optimization/src/ResourceTracker.cpp` `docs/dev/debt/shader.md`
PR 6412	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `docs/dev/debt/shader.md`
PR 6488	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `core/shader/recompiler/Optimization/src/ResourceTracker.cpp` `docs/dev/debt/shader.md`
PR 6924	`docs/dev/debt/shader.md`
PR 7223	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `core/shader/recompiler/Optimization/src/ResourceTracker.cpp` `docs/dev/debt/shader.md`
PR 7392	`docs/dev/debt/shader.md`
PR 7410	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `docs/dev/debt/shader.md`
PR 7642	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `core/shader/recompiler/Optimization/src/ResourceTracker.cpp`
PR 7684	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `core/shader/recompiler/Optimization/src/ResourceTracker.cpp` `docs/dev/debt/shader.md`
PR 7694	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `core/shader/recompiler/Optimization/src/ResourceTracker.cpp`
PR 7922	`docs/dev/debt/shader.md`
PR 8004	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `docs/dev/debt/shader.md`
PR 9020	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `docs/dev/debt/shader.md`
PR 9183	`core/shader/recompiler/Optimization/src/ResourceTracker.cpp` `docs/dev/debt/shader.md`
PR 9335	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `docs/dev/debt/shader.md`
PR 9356	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`
PR 9387	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `docs/dev/debt/shader.md`
PR 9508	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`
PR 9748	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `docs/dev/debt/shader.md`
PR 9852	`core/shader/recompiler/Optimization/src/ResourceTracker.cpp`
PR 10193	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `core/shader/recompiler/Optimization/src/ResourceTracker.cpp`
PR 10233	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `docs/dev/debt/shader.md`
PR 10359	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`
PR 10547	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`
PR 10652	`docs/dev/debt/shader.md`
PR 10757	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`
PR 10867	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `docs/dev/debt/shader.md`
PR 10941	`docs/dev/debt/shader.md`
PR 11351	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`
PR 11483	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `docs/dev/debt/shader.md`
PR 11551	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`
PR 12299	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`
PR 12324	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`
PR 12548	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `core/shader/recompiler/Optimization/src/SrtWalker/SrtEvaluator.cpp` `core/shader/recompiler/Optimization/src/SrtWalker/SrtInstructionPredicates.cpp` `docs/dev/debt/shader.md`
PR 12828	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`
PR 13928	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `core/shader/recompiler/Optimization/src/ResourceTracker.cpp`
PR 13929	`core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` `core/shader/recompiler/Optimization/src/ResourceTracker.cpp`
PR 14212	`docs/dev/debt/shader.md`

oneandonlydean · 2026-10-30T04:50:12Z

@Prophet069 thanks for these. One problem common to all five PRs (#13145, #13146, #13147, #13148, #13149): each commit sits on main 5ecef095b6, but its tree still has the older versions of the files that #12916 and #12923 changed (for example RequestSerializer.cpp, UnclampedNormalizedBlend.cpp, LabelHistory.cpp, WaitMemory.cpp). So each PR, as it stands, would undo both of those merges; that is where the ~465 deleted lines in every diff come from.

For this PR, git diff 08a10d2253 <head> (main just before #12916) shows only your own change. Could you rebase it onto current main keeping only that change (for example git rebase --onto origin/main 08a10d2253 on a branch whose tree is your change applied to 08a10d2253, or re-apply your change on a fresh branch from main), then force-push? We'll review the change itself as soon as the diff no longer touches the #12916/#12923 files.

An image-table offset that is scaled by a constant stride before v_readfirstlane is recognized as a constant-stride column through selects whose condition the active mask implies. Runtime SRT evaluation supports unsigned greater-than-or-equal comparisons, uniform selected components of constructed U32x4 values and known active-mask conjuncts in nested selects. Image dimension queries stay in GPU IR and variable scales are not constant-stride columns. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

Prophet069 · 2026-10-30T05:19:38Z

Thanks, you were right: I squashed onto a moving origin/main, so the commit kept the pre-#12916/#12923 tree. Re-applied only this change on a fresh branch from main 8c8b998620 and force-pushed (same patch-id as git diff 08a10d2253 <old head>); the diff now touches only this PR's files. Rebuilt and reran the focused tests on 8c8b998620 and on the new head with the validation layer; the description has the results.

@MarcZX10 The single-target ExportTargetMerge, the targetMerge field and the Require(merge.mask == 0, ...) came from the stale tree of the first push, which reverted #12916; this PR does not change them. The new head leaves State.cpp, Recompiler.hpp and Recompiler.cpp as on main. agc_driver_unclamped_normalized_blend (2 draws blending two targets each) passes on main 8c8b998620 and on this head with VK_LAYER_KHRONOS_validation, 0 VUID-.

Tehshortbus · 2026-10-30T05:32:19Z

@Prophet069 a heads-up: this branch sets the files #12916 changed back to their contents before #12916 merged (3d2d264392), although its base already includes #12916. For example, core/libs/prx/libSceAgcDriver/tests/execution/UnclampedNormalizedBlend.cpp loses TwoTargetProgram and the two-target draws, and ShaderPixelStageInfo::targetMerges goes back to a single targetMerge. Merged as it is, this would undo #12916, and PPSA26344 would stop again on its two-target blend (#12796). The same happens in #13145, #13146, #13147, #13148 and #13149, so they may share a stale copy of those files.

AI-assisted: yes (Claude Code)

MarcZX10

Re-checked — new head, looks good

The current head preserves the multi-target merge behavior that was missing from the earlier head, and I found no blocking issue in the revised scalarized image-table change.

What I ran

Reviewed head 74952faffa809977ec9b690d38dfbe95f4bdc770 on base 8c8b9986204d5b1cbb23847b363df00ae832ec9c; the merge preview is clean (c2c1c5de3736686f39aace3ea02a5f3ccd05257b). Environment: Linux x86-64. No local build or tests were run.

git diff --check 8c8b9986204d5b1cbb23847b363df00ae832ec9c...74952faffa809977ec9b690d38dfbe95f4bdc770
git merge-tree --write-tree 8c8b9986204d5b1cbb23847b363df00ae832ec9c 74952faffa809977ec9b690d38dfbe95f4bdc770
python3 tools/check_conventions.py --base 8c8b9986204d5b1cbb23847b363df00ae832ec9c --head 74952faffa809977ec9b690d38dfbe95f4bdc770
python3 tools/pr_overlap.py --branch 74952faffa809977ec9b690d38dfbe95f4bdc770 --base main --repo boykopovar/AnyPS5 --jobs 16
gh pr checks 13149 --repo boykopovar/AnyPS5

Results

Check	Result
Diff check	Passed; no output
Merge preview	Clean
Conventions on exact PR head	Passed
Open-PR overlap scan	No confirmed overlapping hunks; 31 comparisons were inconclusive because those candidate PRs conflict with the base. I manually checked the relevant shared files in #12548, #12828 and #12764; their overlapping evaluator/test/documentation changes are in separate cases or entries.
Hosted checks	Windows, Ubuntu/Windows relinker, relinker ASan, conventions, eligibility, labels and report passed. Linux, register-context and write-watch were skipped. Both macOS relinker jobs were pending when checked.
Author-reported focused tests	63/63 on main and the new head, plus the two-target blend regression on both; same Windows/RTX 3080 environment, no VUIDs reported. Not independently run.
Local build/tests	Not run; no build performed

Unedited output

git diff --check 8c8b9986204d5b1cbb23847b363df00ae832ec9c...74952faffa809977ec9b690d38dfbe95f4bdc770
(no output; exit status 0)

git merge-tree --write-tree 8c8b9986204d5b1cbb23847b363df00ae832ec9c 74952faffa809977ec9b690d38dfbe95f4bdc770
c2c1c5de3736686f39aace3ea02a5f3ccd05257b

python3 tools/check_conventions.py --base 8c8b9986204d5b1cbb23847b363df00ae832ec9c --head 74952faffa809977ec9b690d38dfbe95f4bdc770
All conventions checks passed.

python3 tools/pr_overlap.py --branch 74952faffa809977ec9b690d38dfbe95f4bdc770 --base main --repo boykopovar/AnyPS5 --jobs 16
<!-- pr-overlap -->
Some comparisons could not be completed because a pull request conflicts with the base branch. Review these shared files manually before concluding that there is no overlap.

| Pull request | Shared files not compared |
|---|---|
| [PR 5021](https://www.github.com/boykopovar/AnyPS5/pull/5021) | `docs/dev/debt/shader.md` |
| [PR 5371](https://www.github.com/boykopovar/AnyPS5/pull/5371) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`<br>`docs/dev/debt/shader.md` |
| [PR 5609](https://www.github.com/boykopovar/AnyPS5/pull/5609) | `core/shader/recompiler/Optimization/src/SrtWalker/SrtEvaluator.cpp` |
| [PR 6235](https://www.github.com/boykopovar/AnyPS5/pull/6235) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`<br>`docs/dev/debt/shader.md` |
| [PR 6280](https://www.github.com/boykopovar/AnyPS5/pull/6280) | `core/shader/recompiler/Optimization/src/ResourceTracker.cpp`<br>`docs/dev/debt/shader.md` |
| [PR 6412](https://www.github.com/boykopovar/AnyPS5/pull/6412) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`<br>`docs/dev/debt/shader.md` |
| [PR 6488](https://www.github.com/boykopovar/AnyPS5/pull/6488) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`<br>`core/shader/recompiler/Optimization/src/ResourceTracker.cpp`<br>`docs/dev/debt/shader.md` |
| [PR 6924](https://www.github.com/boykopovar/AnyPS5/pull/6924) | `docs/dev/debt/shader.md` |
| [PR 7392](https://www.github.com/boykopovar/AnyPS5/pull/7392) | `docs/dev/debt/shader.md` |
| [PR 7642](https://www.github.com/boykopovar/AnyPS5/pull/7642) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`<br>`core/shader/recompiler/Optimization/src/ResourceTracker.cpp` |
| [PR 7684](https://www.github.com/boykopovar/AnyPS5/pull/7684) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`<br>`core/shader/recompiler/Optimization/src/ResourceTracker.cpp`<br>`docs/dev/debt/shader.md` |
| [PR 7922](https://www.github.com/boykopovar/AnyPS5/pull/7922) | `docs/dev/debt/shader.md` |
| [PR 8004](https://www.github.com/boykopovar/AnyPS5/pull/8004) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`<br>`docs/dev/debt/shader.md` |
| [PR 9020](https://www.github.com/boykopovar/AnyPS5/pull/9020) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`<br>`docs/dev/debt/shader.md` |
| [PR 9335](https://www.github.com/boykopovar/AnyPS5/pull/9335) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`<br>`docs/dev/debt/shader.md` |
| [PR 9356](https://www.github.com/boykopovar/AnyPS5/pull/9356) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` |
| [PR 9387](https://www.github.com/boykopovar/AnyPS5/pull/9387) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`<br>`docs/dev/debt/shader.md` |
| [PR 9748](https://www.github.com/boykopovar/AnyPS5/pull/9748) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`<br>`docs/dev/debt/shader.md` |
| [PR 9889](https://www.github.com/boykopovar/AnyPS5/pull/9889) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`<br>`core/shader/recompiler/Optimization/src/ResourceTracker.cpp`<br>`docs/dev/debt/shader.md` |
| [PR 9897](https://www.github.com/boykopovar/AnyPS5/pull/9897) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`<br>`core/shader/recompiler/Optimization/src/ResourceTracker.cpp` |
| [PR 10193](https://www.github.com/boykopovar/AnyPS5/pull/10193) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`<br>`core/shader/recompiler/Optimization/src/ResourceTracker.cpp` |
| [PR 10233](https://www.github.com/boykopovar/AnyPS5/pull/10233) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`<br>`docs/dev/debt/shader.md` |
| [PR 10547](https://www.github.com/boykopovar/AnyPS5/pull/10547) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` |
| [PR 10652](https://www.github.com/boykopovar/AnyPS5/pull/10652) | `docs/dev/debt/shader.md` |
| [PR 10757](https://www.github.com/boykopovar/AnyPS5/pull/10757) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` |
| [PR 11351](https://www.github.com/boykopovar/AnyPS5/pull/11351) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` |
| [PR 11551](https://www.github.com/boykopovar/AnyPS5/pull/11551) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` |
| [PR 12324](https://www.github.com/boykopovar/AnyPS5/pull/12324) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` |
| [PR 12548](https://www.github.com/boykopovar/AnyPS5/pull/12548) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp`<br>`core/shader/recompiler/Optimization/src/SrtWalker/SrtEvaluator.cpp`<br>`core/shader/recompiler/Optimization/src/SrtWalker/SrtInstructionPredicates.cpp`<br>`docs/dev/debt/shader.md` |
| [PR 12764](https://www.github.com/boykopovar/AnyPS5/pull/12764) | `docs/dev/debt/shader.md` |
| [PR 12828](https://www.github.com/boykopovar/AnyPS5/pull/12828) | `core/libs/prx/libSceAgcDriver/tests/ShaderMemory.cpp` |


gh pr checks 13149 --repo boykopovar/AnyPS5
linux\tskipping\t0\thttps://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906528372
register-context\tskipping\t0\thttps://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906527454
relinker (ubuntu-latest, g++)\tpass\t2m43s\thttps://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906526464
relinker (windows-latest, g++)\tpass\t7m20s\thttps://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906526407
windows\tpass\t4m47s\thttps://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906526357
write-watch\tskipping\t0\thttps://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906527937
relinker-asan\tpass\t2m41s\thttps://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906526436
relinker (macos-latest, clang++)\tpending\t0\thttps://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906526467
relinker (macos-latest, clang++, x86_64)\tpending\t0\thttps://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906526413
conventions\tpass\t13s\thttps://github.com/boykopovar/AnyPS5/actions/runs/41007774250/job/123906463794
eligibility\tpass\t7s\thttps://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906464931
labels\tpass\t7s\thttps://github.com/boykopovar/AnyPS5/actions/runs/41007774317/job/123906502619
report\tpass\t51s\thttps://github.com/boykopovar/AnyPS5/actions/runs/41007775466/job/123906463951

Findings

Blocking

None on the current head. The earlier multi-target finding applied to the stale head 6bf492e804cefd0ef70ab4f4a062aeaa5eeed02d; this head has no changes to State.cpp or the recompiler merge-state files, and the author reports that the two-target regression test passes on main and this branch.

Non-blocking

No further issue found in the reviewed changes. Linux product/specialized jobs were skipped, macOS relinker checks were pending, and GPU-focused tests were not independently run here.

Sources

Not verified

I did not run a local build, CTest, Linux product tests, console behavior, or GPU validation. The 63/63 tests and the two-target blend test are the author's report, not a local run. The two macOS relinker jobs were still pending at the CI check snapshot.

MarcZX10 · 2026-10-30T05:48:52Z

Formatting correction — CI output

The review is approved on the current head; I corrected the collapsed check excerpt because its separators were shown as literal text instead of tabs.

linux	skipping	0	https://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906528372
register-context	skipping	0	https://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906527454
relinker (ubuntu-latest, g++)	pass	2m43s	https://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906526464
relinker (windows-latest, g++)	pass	7m20s	https://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906526407
windows	pass	4m47s	https://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906526357
write-watch	skipping	0	https://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906527937
relinker-asan	pass	2m41s	https://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906526436
relinker (macos-latest, clang++)	pending	0	https://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906526467
relinker (macos-latest, clang++, x86_64)	pending	0	https://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906526413
conventions	pass	13s	https://github.com/boykopovar/AnyPS5/actions/runs/41007774250/job/123906463794
eligibility	pass	7s	https://github.com/boykopovar/AnyPS5/actions/runs/41007774550/job/123906464931
labels	pass	7s	https://github.com/boykopovar/AnyPS5/actions/runs/41007774317/job/123906502619
report	pass	51s	https://github.com/boykopovar/AnyPS5/actions/runs/41007775466/job/123906463951

TechLevelZero · 2026-10-30T12:16:42Z

✅ NVIDIA GPU test: passed

PR head 74952fa merged onto main 73f4f7e · tested 2026-10-31 20:33 UTC · 33 min


GPU	NVIDIA RTX 2000 Ada Generation, driver 595.99.02, Vulkan 1.4.329
Host	Ubuntu 26.04.1 LTS, kernel 7.0.0-38-generic, container ubuntu:24.04
Logs	build and test logs for this run
Build	✅ passed (Release, `-DANYPS5_ENABLE_SPIRV_TOOLS=ON`)
Tests	✅ 2,336 passed · 4 failed (0 new, 3 also fail on main, 1 unstable on main) · 48 skipped (`ANYPS5_REQUIRE_VULKAN=1`)
Test setup	CPU-only tests without a GPU; GPU tests (and any CPU-only test that failed or skipped without one) once on the GPU with `VK_LAYER_KHRONOS_validation` and `ANYPS5_NO_SHADER_CACHE=1`; main ran the same way
Test selection	2,388 of 3,072 tests run: the ones this change can reach, chosen by `tools/select_tests.py` from main's build graph
Vulkan validation	✅ no new validation IDs vs main (878 tests run on the GPU)
Slow tests (>30 s, GPU tests timed with validation layers)	⚠️ `agc_driver_graphics_maps_fallback` 30.1 s

Unstable on main (1, these also failed in recent main runs on this machine; not counted against this PR)

agc_driver_submit_progress: failed in 8 of the last 10 main runs

Failing on main too (3, not counted against this PR)

agc_submit_progress
guest_mmap_executable
guest_socket_options_multicast

Validation IDs already present on main (3 IDs in 387 tests)

VUID-VkDeviceCreateInfo-pNext-pNext (387 tests): agc_compute_pipeline_replay, agc_driver, agc_driver_address_passes, agc_driver_address_passes_draw_transitions, agc_driver_address_passes_targets_split, agc_driver_alias_copy_label, agc_driver_alpha_to_coverage, agc_driver_array_layer_target, and 379 more
VUID-vkCmdCopyImage-srcImage-01548 (12 tests): agc_driver_depth_clear_hle, agc_driver_depth_plane_buffer_memory, agc_driver_depth_plane_cpu_write_after_memo, agc_driver_depth_plane_cpu_write_after_page_unwatch, agc_driver_depth_plane_cpu_write_after_partial_unwatch, agc_driver_depth_plane_partial_store, agc_driver_depth_plane_partial_store_unimported, agc_driver_depth_plane_sampling, and 4 more
VUID-vkCmdCopyImage-srcImage-01551 (12 tests): agc_driver_depth_clear_hle, agc_driver_depth_plane_buffer_memory, agc_driver_depth_plane_cpu_write_after_memo, agc_driver_depth_plane_cpu_write_after_page_unwatch, agc_driver_depth_plane_cpu_write_after_partial_unwatch, agc_driver_depth_plane_partial_store, agc_driver_depth_plane_partial_store_unimported, agc_driver_depth_plane_sampling, and 4 more

Skipped (48)

agc_driver_bda_reads_shadowed_results: skipped, the device reports watched host memory written when it imports or reads it
agc_driver_bda_reads_shadowed_results_per_build: skipped, the device reports watched host memory written when it imports or reads it
agc_driver_msaa_dcc_compared_import_2: skipped, the device reports host imports that keep write protection, so the write-watch probe compares none of them
agc_driver_msaa_dcc_compared_import_4: skipped, the device reports host imports that keep write protection, so the write-watch probe compares none of them
agc_driver_msaa_dcc_compared_import_8: skipped, the device reports host imports that keep write protection, so the write-watch probe compares none of them
agc_driver_sample_order_grid_reference: skipped, the device does not keep depth drawn at custom sample locations at those locations, even after a transition that names them: the composite of uniform renders holds the guest's depth in 0 of 3005 sloped samples, so no path can be compared with it bit for bit
agc_driver_vertex_elected_atomic_max: skipped, the device runs vertex programs in subgroups, so they keep their cross-lane reductions
agc_host_import_sparse: skipped, host imports are unavailable: /dev/udmabuf is not accessible
agc_import_settle_initial_dcc: skipped, the device has no deferred settle of host imports: imports under 1 MiB wait for one only on Windows
agc_unit_shadow_completion_store: skipped, the device reports watched host memory written when it imports or reads it
guest_http_uri_locale: 
guest_net_ctl: skipped, the host has no IPv4 address
guest_video_out_color_space: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_hdr: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_high_frame_rate: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_output_mode: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_resolution_status: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_wqhd_detection: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
video_out_flip_aftergpu: skipped, no display or Vulkan device: DisplayWindow: Vulkan support is either not configured in SDL or not available in current SDL video driver (offscreen) or platform
video_out_flip_backtoback: skipped, no display or Vulkan device: DisplayWindow: Vulkan support is either not configured in SDL or not available in current SDL video driver (offscreen) or platform
video_out_flip_controls: skipped, no display or Vulkan device: DisplayWindow: Vulkan support is either not configured in SDL or not available in current SDL video driver (offscreen) or platform
video_out_flip_cpuflip: skipped, no display or Vulkan device: DisplayWindow: Vulkan support is either not configured in SDL or not available in current SDL video driver (offscreen) or platform
video_out_flip_cpuflip_in_order: skipped, no display or Vulkan device: DisplayWindow: Vulkan support is either not configured in SDL or not available in current SDL video driver (offscreen) or platform
video_out_flip_dcc: skipped, no display or Vulkan device: DisplayWindow: Vulkan support is either not configured in SDL or not available in current SDL video driver (offscreen) or platform
video_out_flip_deferred: skipped, no display or Vulkan device: DisplayWindow: Vulkan support is either not configured in SDL or not available in current SDL video driver (offscreen) or platform
video_out_flip_eop: skipped, no display or Vulkan device: DisplayWindow: Vulkan support is either not configured in SDL or not available in current SDL video driver (offscreen) or plat

Reproduce

git fetch origin 74952faffa809977ec9b690d38dfbe95f4bdc770 73f4f7e06e4f210a8c6570e335fd654dc0a86134
git checkout 73f4f7e06e4f210a8c6570e335fd654dc0a86134 && git merge --no-ff 74952faffa809977ec9b690d38dfbe95f4bdc770
git submodule update --init --recursive
cmake -S . -B build -G Ninja -DCMAKE_BUILD_TYPE=Release -DBUILD_TESTING=ON -DANYPS5_ENABLE_SPIRV_TOOLS=ON
cmake --build build && cmake --build build --target libs
VK_DRIVER_FILES=/nonexistent/icd.json ctest --test-dir build -E '^(agc|shader|recompiler|videoout|video_out|guest_video_out|vulkan)|gpu' --timeout 120 --parallel 32 --output-on-failure
ANYPS5_REQUIRE_VULKAN=1 ANYPS5_NO_SHADER_CACHE=1 VK_INSTANCE_LAYERS=VK_LAYER_KHRONOS_validation \
  ctest --test-dir build -R '^(agc|shader|recompiler|videoout|video_out|guest_video_out|vulkan)|gpu' --timeout 120 --parallel 4 --output-on-failure
# failures are then re-run alone: --timeout 360 --parallel 2; new validation IDs: the test alone, twice, --parallel 1

_{Automated build and test on a contributor's NVIDIA machine; it does not replace upstream CI. A new comment is posted for each new push. Full logs.}

RNK123 · 2026-11-01T03:25:02Z

Independent Linux / AMD RADV focused validation of head 74952faffa809977ec9b690d38dfbe95f4bdc770, merged without conflicts onto main dc9dc7eaaeaf99b1c491d426e68bb999def3da16. Tested merged tree: 99a81bef7a2a15111f15a4f16c60046651625c9a (temporary uncommitted merge; no source changes added).

Host: CachyOS Linux 7.3.1-1-cachyos, Intel Core i5-12400F, GCC 16.2.1, Release, BUILD_TESTING=ON, CTest 4.4.5. GPU: AMD Radeon RX 9060 XT (1002:7590, RADV GFX1200, subgroup 64), Mesa 26.2.5-arch3.1, Vulkan 1.4.354. Khronos validation layer 1.4.363.

Method: Same machine, build directory and exact CTest selection on both trees, serial --parallel 1 --repeat until-fail:3 --timeout 60 -V, with ANYPS5_REQUIRE_VULKAN=1, ANYPS5_NO_SHADER_CACHE=1 and VK_INSTANCE_LAYERS=VK_LAYER_KHRONOS_validation. The locally installed layer was exposed through VK_LAYER_PATH; VK_LOADER_DEBUG=layer confirmed its insertion in both logs (15 insertions each). Neither log contains was not found but was requested by env var VK_INSTANCE_LAYERS. GPU selection messages identify the Radeon above, not a software device.

Tree	Focused entries passed	Individual executions passed	Skipped / failed	`VUID-` / `SYNC-HAZARD-`
`main` `dc9dc7eaae`	66/66, each 3/3	198/198	0 / 0	0 / 0
Same `main` + head `74952faffa`	66/66, each 3/3	198/198	0 / 0	0 / 0

The selection covers all entries invoking agc_shader_memory_tests (21), agc_driver_cmask_fast_clear_tests (1), agc_driver_graphics_tests (40), agc_driver_sample_multisample_target_tests (3), plus agc_driver_unclamped_normalized_blend_tests (1). These are the four executable families named in the description and the separate multi-target regression, enumerated on the newer main, hence 66 entries rather than the earlier 63 plus one. The PR's added shader-memory checks run as part of agc_shader_memory.

Existing warnings, not hidden: Both logs contain 30 Undefined-Value-ShaderOutputNotConsumed-DynamicRendering validation warnings, all from agc_driver_unclamped_normalized_blend (10 per execution): fragment output location 0 is unused when color attachment 0 is null. Warning identifiers and counts match; no validation errors or new warning identifiers were observed. Both trees report 2 draws blending two targets each on all three executions of that regression.

Scope: This is focused regression/build coverage, including real GPU rendering tests, not a game bring-up or benchmark. Some entries, including the new runtime-value checks, are CPU-side checks. I did not obtain or replay the captured PPSA02015 shader, run the game, or independently establish live GPU behavior of the new scalarized image-table path; the PR's technical-debt caveat remains applicable. No console-equivalence claim.

Equivalent reproducible selection

After building the five targets named above, generate a CTest name list and run it on each tree with the environment described above:

python3 - <<'PY'
import json
import subprocess
from pathlib import Path

targets = {
    "agc_shader_memory_tests",
    "agc_driver_cmask_fast_clear_tests",
    "agc_driver_graphics_tests",
    "agc_driver_sample_multisample_target_tests",
    "agc_driver_unclamped_normalized_blend_tests",
}
data = json.loads(subprocess.check_output(
    ["ctest", "--test-dir", "build-main", "--show-only=json-v1"], text=True))
names = [test["name"] for test in data["tests"]
         if Path(test.get("command", [""])[0]).name in targets]
assert len(names) == 66
Path("radv-focused-tests.txt").write_text("\n".join(names) + "\n")
PY
ctest --test-dir build-main --tests-from-file "$PWD/radv-focused-tests.txt" \
  --parallel 1 --repeat until-fail:3 --timeout 60 --no-tests=error -V

_{AI-assisted hardware test run (Codex). No additional PR, code changes, or performance claim.}

github-actions Bot added the unblocks title Names a title ID on the Unblocks line label Oct 30, 2026

MarcZX10 suggested changes Oct 30, 2026

View reviewed changes

oneandonlydean mentioned this pull request Oct 30, 2026

feat(agc): report live device memory allocations when the device is lost #13148

Open

12 tasks done

Prophet069 force-pushed the pr/shader-scalarized-image-tables branch from 6bf492e to 74952fa Compare October 30, 2026 05:19

Tehshortbus mentioned this pull request Oct 30, 2026

fix(agc): look up a merged color target again when a recovery evicted its resident image #13178

Merged

12 tasks done

MarcZX10 approved these changes Oct 30, 2026

View reviewed changes

oneandonlydean self-assigned this Oct 30, 2026

github-actions Bot added shader core/shader docs agc libSceAgc* labels Oct 30, 2026

oneandonlydean merged commit b5f7270 into boykopovar:main Nov 1, 2026
14 of 15 checks passed

hugodearaujo mentioned this pull request Nov 2, 2026

fix(shader): plan several T#s read through one pointer-loaded V#, keeping the V# real for their other uses #9790

Open

12 tasks done

Navigation Menu

Repository navigation

fix(shader): recognize scalarized constant-stride image tables - #13149

fix(shader): recognize scalarized constant-stride image tables#13149
oneandonlydean merged 1 commit into
boykopovar:mainfrom
Prophet069:pr/shader-scalarized-image-tables

Prophet069 commented Oct 30, 2026 •

edited

Loading

Uh oh!

MarcZX10 left a comment •

edited

Loading

Uh oh!

github-actions Bot commented Oct 30, 2026 •

edited

Loading

Uh oh!

oneandonlydean commented Oct 30, 2026

Uh oh!

Prophet069 commented Oct 30, 2026

Uh oh!

Tehshortbus commented Oct 30, 2026

Uh oh!

MarcZX10 left a comment

Uh oh!

MarcZX10 commented Oct 30, 2026

Uh oh!

TechLevelZero commented Oct 30, 2026 •

edited

Loading

Uh oh!

RNK123 commented Nov 1, 2026

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

Conversation

Prophet069 commented Oct 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What

Tested

Checklist

Uh oh!

MarcZX10 left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Reviewed — changes requested

What I ran

Results

Negative control

Findings

Blocking

Non-blocking

Sources

Not verified

To get this to a good state

Uh oh!

github-actions Bot commented Oct 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

oneandonlydean commented Oct 30, 2026

Uh oh!

Prophet069 commented Oct 30, 2026

Uh oh!

Tehshortbus commented Oct 30, 2026

Uh oh!

MarcZX10 left a comment

Choose a reason for hiding this comment

Re-checked — new head, looks good

What I ran

Results

Findings

Blocking

Non-blocking

Sources

Not verified

Uh oh!

MarcZX10 commented Oct 30, 2026

Formatting correction — CI output

Uh oh!

TechLevelZero commented Oct 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

✅ NVIDIA GPU test: passed

Uh oh!

RNK123 commented Nov 1, 2026

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

Prophet069 commented Oct 30, 2026 •

edited

Loading

MarcZX10 left a comment •

edited

Loading

github-actions Bot commented Oct 30, 2026 •

edited

Loading

TechLevelZero commented Oct 30, 2026 •

edited

Loading