fix(agc): import a depth plane aliased by a cleared R32F target by edubox · Pull Request #15443 · boykopovar/AnyPS5

edubox · 2026-11-03T10:12:23Z

What

A native multisample depth plane whose guest range exactly aliases a multisampled R32F colour target can now take the value that target's clear or fill proves, instead of refusing the depth import. This covers both import paths:

Without HTILE (the uniform-plane path), the depth value comes from the R32F clear instead of guest memory, whose samples are only on the host.
With HTILE, the metadata must prove the same value; a colour clear that disagrees with the HTILE is still refused.

A colour target that has drawn, a non-R32F format, or a clear value outside [0, 1] is still refused. CheckCurrent repeats the check, so an imported depth stays valid only while the colour alias still proves the imported value.

7th Guest (PPSA17019) previously stopped at native multisample depth or HTILE aliases a multisampled colour target whose samples are only on the host; with this it advances to the next, separate blocker (overlapping native multisample depth clear transitions are not implemented).

Tested

Linux x86-64 (CachyOS), GCC 16.2.1, Release, NVIDIA GeForce RTX 4080 SUPER (driver 615.78.8.0):

New agc_driver_native_depth_uniform_plane_alias and agc_driver_native_depth_htile_plane_alias cases: clears 0 and 1.0, a drawn target, and a clear outside [0, 1], with 2, 4 and 8 samples.
ANYPS5_REQUIRE_VULKAN=1 ctest --test-dir build -R 'agc_driver_native_depth|^agc_driver_graphics$|^agc_driver_key_miss$|^agc_driver_bda_device$|^agc_driver_recorder$': 19/19 pass on the change, 17/17 on main (4f41f4c375).
VK_LAYER_KHRONOS_validation is not installed on this host, so the validation-layer comparison could not be run.
tools/check_conventions.py --base origin/main and git diff --check are clean.
tools/pr_overlap.py --branch fix/agc-aliased-uniform-depth flags a shared DepthSurface.cpp with fix(agc): keep guest writes over a stale stencil plane #14969 and perf(agc): size thin surfaces without a heap mip chain #15059; reviewed manually: they touch stencil retirement/flush and surface-size helpers, not this import or check path.

Checklist

oneandonlydean · 2026-11-03T10:20:41Z

Hi edubox,

Thanks for this; the alias rules (R32F only, a proving clear or fill, a value in [0, 1], an exact range, and CheckCurrent re-proving it) read right. Three points on 12006953513, the first being the one that matters most:

The depth-alias proof should stay separate from the stencil-fill proof in DepthSurface::CheckCurrent. With this change, an accepted alias at index zero sets planeAliased = true. In the later stencilClearGeneration != 0 branch that same flag sets sampling.aliasStencil, bypasses the complete tracked-store requirement and skips checking the stencil bytes. So a valid R32F depth clear is enough even when the separate stencil plane has no alias proving its fill. Could you track the stencil alias independently and keep the existing stencil store/value checks unless that stencil plane itself proves the value? A regression that combines a valid depth alias with a stencil fill whose tracked store or value is missing, next to the successful combined depth/stencil case, would cover it. (This is from reading the source, not a reproduced GPU failure.)
The new tests pin a refusal for a valid state. In CheckUniformPlaneAliasedDepth and CheckHtileAliasedDepth, the step that views the colour target once more and calls TargetRefuses(..., depthAliasRefusal) requires a throw for a colour target that has drawn. That is valid; the importer just doesn't support it yet, so whoever adds support would have to edit this test first. Could you drop that refusal check from the new cases? The accepted-path checks, the out-of-range clear and the HTILE-disagrees case don't need it.
One HTILE combination is accepted but untested. prepareDepthSampling proves the HTILE words against state.aliasDepth.value_or(target.clearDepth), and the import clears with the alias value. So an HTILE that agrees with the colour clear is accepted even when the target's own depth clear differs (for example, the target clears to 0 while the colour clear and HTILE say 1). The HTILE test only covers 0/0. If that's intended, a case for it would show the sampled value; if not, the alias value should also have to equal target.clearDepth.

edubox · 2026-11-03T10:48:15Z

Addressed all three points in 6e9b2bee4c:

CheckCurrent now tracks the stencil plane alias separately (stencilPlaneAliased). A depth alias no longer marks the separate stencil plane as aliased or bypasses its tracked-store/value checks. Added the combined regression: a valid depth alias with a stored stencil fill is accepted, and the same fill without its store is refused.
Removed the drawn-colour-target refusal from both new tests; the accepted paths, the out-of-range clear and the HTILE-disagrees case remain.
Kept the HTILE proof against the alias value, and added a case where target.clearDepth is 0 while the colour clear and HTILE say 1, with the sampled value checked as 1.

Focused tests: 19/19 (agc_driver_native_depth*, agc_driver_graphics, agc_driver_key_miss, agc_driver_bda_device, agc_driver_recorder). check_conventions.py --base origin/main and git diff --check are clean.

RNK123 · 2026-11-03T11:20:19Z

Independent RADV run: both new depth-alias tests pass on this head, and the same tests reproduce the prior refusal when the production file is temporarily returned to main's version.

Linux (CachyOS kernel 7.3.1-1-cachyos), GCC 16.2.1 Release, Radeon RX 9060 XT (RADV GFX1200, Mesa 26.2.5, Vulkan 1.4.354), Khronos validation layer 1.4.363. Head 6e9b2bee4ca544a1cc2569092bbe36c60efc754b was merged locally onto main 4f41f4c375f5abb487dbd184fbc150f31994d762 (tree 88f041445051fbe2e5b5752b44f08fbdd0c52c58).

Three existing controls (agc_driver_native_depth_separate_stencil, _aliased_stencil, _uniform_planes) passed 3/3 each on main and the merged candidate.
Both new controls (agc_driver_native_depth_uniform_plane_alias, _htile_plane_alias) passed 3/3 each on the candidate. Each reports the 2-, 4-, and 8-sample cases checked and zero unavailable cases.
Counter-test on the merged stack: retaining the new test files but temporarily replacing only DepthSurface.cpp with its byte-identical main version made each new test fail 3/3 with the prior native multisample depth or HTILE aliases a multisampled colour target whose samples are only on the host refusal. Restoring the candidate file made both pass 3/3. This establishes sensitivity to that refusal, not a reproduction of a game issue.
The runs selected the RX 9060 XT and loaded the requested Khronos layer. No test or sample-count case was skipped and no VUID- or SYNC-HAZARD- appeared in these focused logs. The unrelated LSFG implicit layer reported a loader error in both arms; the Khronos layer was inserted.

Commands

export ANYPS5_REQUIRE_VULKAN=1 ANYPS5_NO_SHADER_CACHE=1
export VK_INSTANCE_LAYERS=VK_LAYER_KHRONOS_validation VK_LOADER_DEBUG=layer
export VK_LAYER_PATH="$HOME/.local/share/anyps5-tester/vk-layers/usr/share/vulkan/explicit_layer.d"
cmake --build build-main --target agc_driver_native_multisample_depth_tests -j 4
ctest --test-dir build-main -R '^agc_driver_native_depth_(uniform_planes|aliased_stencil|separate_stencil)$' --repeat until-fail:3 -V
ctest --test-dir build-main -R '^agc_driver_native_depth_(uniform_plane_alias|htile_plane_alias)$' --repeat until-fail:3 -V

The last command requires the candidate test files; the counter-test additionally substitutes main's DepthSurface.cpp locally and restores it afterwards.

No game, performance claim, or other GPU/driver was tested. AI-assisted independent hardware test run (GPT-6.1 Terra medium).

TechLevelZero · 2026-11-03T11:50:04Z

✅ AMD GPU test: passed

PR head 6e9b2be merged onto main 4f41f4c · tested 2026-11-03 11:50 UTC · 17 min


GPU	AMD Radeon RX 6300 (RADV NAVI24), driver Mesa 25.2.8-0ubuntu0.24.04.4, Vulkan 1.4.318
Host	Ubuntu 26.04.1 LTS, kernel 7.0.0-42-generic, container ubuntu:24.04
Logs	build and test logs for this run
Build	✅ passed (Release, `-DANYPS5_ENABLE_SPIRV_TOOLS=ON`)
Tests	✅ 2,733 passed · 10 failed (0 new, 8 also fail on main, 2 unstable on main) · 58 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	2,801 of 3,811 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 (925 tests run on the GPU)
Slow tests (>30 s, GPU tests timed with validation layers)	⚠️ `agc_driver_pipeline_journal` 120.4 s, ⚠️ `agc_driver_image_sample_unnormalized` 62.0 s, ⚠️ `public_headers` 59.6 s, ⚠️ `agc_driver_lazy_resolved_abi` 47.2 s, ⚠️ `agc_shader_sampler_limits` 36.1 s

Unstable on main (2, these also failed in recent main runs on this machine; not counted against this PR)

agc_driver_pipeline_journal: failed in 7 of the last 9 main runs
agc_driver_store_over_untracked: failed in 1 of the last 9 main runs

Failing on main too (8, not counted against this PR)

agc_driver_image_table
agc_driver_native_depth_bda_clear
agc_driver_native_multisample_depth
agc_driver_submit_progress
agc_submit_progress
agc_submit_progress_no_timeline
guest_mmap_executable
guest_socket_options_multicast

Validation IDs already present on main (4 IDs in 517 tests)

VUID-VkDeviceCreateInfo-pNext-pNext (517 tests): agc_driver, agc_driver_address_passes, agc_driver_alias_copy_label, agc_driver_alpha_to_coverage, agc_driver_array_layer_target, agc_driver_bc_block_copy, agc_driver_bda_anchor_reads, agc_driver_bda_span_reads, and 509 more
VUID-vkCmdCopyImage-srcImage-01548 (8 tests): agc_driver_depth_clear_hle, agc_driver_depth_plane_buffer_memory, agc_driver_depth_plane_partial_store, agc_driver_depth_storage_dispatch, agc_driver_depth_surface_from_storage, agc_driver_dispatch_move_depth, agc_driver_stencil_null_feedback, agc_driver_stencil_plane_memory
VUID-vkCmdCopyImage-srcImage-01551 (8 tests): agc_driver_depth_clear_hle, agc_driver_depth_plane_buffer_memory, agc_driver_depth_plane_partial_store, agc_driver_depth_storage_dispatch, agc_driver_depth_surface_from_storage, agc_driver_dispatch_move_depth, agc_driver_stencil_null_feedback, agc_driver_stencil_plane_memory
VUID-VkPipelineColorBlendStateCreateInfo-pAttachments-00605 (1 test): agc_driver_srgb8_color_target

Skipped (58)

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_copied_writer_label_order: skipped, the device reports a minStorageBufferOffsetAlignment of 4, so the dispatch binds its record at +0x1004 in place and nothing is written back; the test needs an alignment above 4
agc_driver_float16_unused_nan_payloads: skipped, the device has no native f16 conversions with RTE rounding and kept f16 denormals
agc_driver_image_table_vouched_lookups: skipped, the device reports 4 of 4 host-imported pages written after a GPU read, so a watched imported table cannot be vouched for
agc_driver_linear_target_direct_write_completion: skipped, the device does not keep host imports write-watched, so no completion stamp over the target is tracked
agc_driver_ngg_vertex_wave_reduction: skipped, s3 describes a 32-lane wave only on 32-lane subgroups, and the device has 64 lanes
agc_driver_scattered_runs: skipped, host imports are unavailable: /dev/udmabuf is not accessible
agc_driver_sparse_dma_buf_import: skipped, host imports are unavailable: /dev/udmabuf is not accessible
agc_driver_unit_shadow: skipped, the device does not import watched host memory for unit shadows
agc_driver_unit_shadow_budget: skipped, the device does not import watched host memory for unit shadows
agc_host_import_sparse: skipped, host imports are unavailable: /dev/udmabuf is not accessible
agc_idle_publish_wait: skipped, the device reports 4 of 4 host-imported pages written after a GPU read, so a forced watch of the import reads its submissions as CPU stores
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_shadowed_key_stores: skipped, the device imports no watched host memory under a unit shadow
agc_small_tiled_shadow_label: skipped, host imports are unavailable: the test block is not imported and watched
agc_storage_clear_hand_over: skipped, the device imports no watched host memory for the clear-key hand-over
agc_unit_shadow: skipped, the device does not import watched host memory for unit shadows
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: DisplayW

Reproduce

git fetch origin 6e9b2bee4ca544a1cc2569092bbe36c60efc754b 4f41f4c375f5abb487dbd184fbc150f31994d762
git checkout 4f41f4c375f5abb487dbd184fbc150f31994d762 && git merge --no-ff 6e9b2bee4ca544a1cc2569092bbe36c60efc754b
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 12 --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 16 --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 AMD machine; it does not replace upstream CI. A new comment is posted for each new push. Full logs.}

TechLevelZero · 2026-11-03T12:04:09Z

✅ NVIDIA GPU test: passed

PR head 6e9b2be merged onto main 4f41f4c · tested 2026-11-03 12:04 UTC · 25 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	✅ 2,748 passed · 5 failed (0 new, 3 also fail on main, 2 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	2,801 of 3,811 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 (925 tests run on the GPU)
Slow tests (>30 s, GPU tests timed with validation layers)	⚠️ `agc_driver_pipeline_journal` 120.1 s

Unstable on main (2, these also failed in recent main runs on this machine; not counted against this PR)

agc_driver_pipeline_journal: failed in 4 of the last 10 main runs
agc_submit_progress_no_timeline: failed in 5 of the last 10 main runs

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

agc_driver_image_table
guest_mmap_executable
guest_socket_options_multicast

Validation IDs already present on main (4 IDs in 517 tests)

VUID-VkDeviceCreateInfo-pNext-pNext (517 tests): agc_driver, agc_driver_address_passes, agc_driver_alias_copy_label, agc_driver_alpha_to_coverage, agc_driver_array_layer_target, agc_driver_bc_block_copy, agc_driver_bda_anchor_reads, agc_driver_bda_span_reads, and 509 more
VUID-vkCmdCopyImage-srcImage-01548 (9 tests): agc_driver_depth_clear_hle, agc_driver_depth_plane_buffer_memory, agc_driver_depth_plane_partial_store, agc_driver_depth_plane_sampling, agc_driver_depth_storage_dispatch, agc_driver_depth_surface_from_storage, agc_driver_dispatch_move_depth, agc_driver_stencil_null_feedback, and 1 more
VUID-vkCmdCopyImage-srcImage-01551 (9 tests): agc_driver_depth_clear_hle, agc_driver_depth_plane_buffer_memory, agc_driver_depth_plane_partial_store, agc_driver_depth_plane_sampling, agc_driver_depth_storage_dispatch, agc_driver_depth_surface_from_storage, agc_driver_dispatch_move_depth, agc_driver_stencil_null_feedback, and 1 more
VUID-VkPipelineColorBlendStateCreateInfo-pAttachments-00605 (1 test): agc_driver_srgb8_color_target

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_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_scattered_runs: skipped, host imports are unavailable: /dev/udmabuf is not accessible
agc_driver_sparse_dma_buf_import: skipped, host imports are unavailable: /dev/udmabuf is not accessible
agc_driver_stencil_ref_export: skipped, the device lacks VK_EXT_shader_stencil_export
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 platform
video_out_flip_eop_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_eop_main: skipped, no display or Vulkan device: DisplayWindow: Vulkan support is either not co

Reproduce

git fetch origin 6e9b2bee4ca544a1cc2569092bbe36c60efc754b 4f41f4c375f5abb487dbd184fbc150f31994d762
git checkout 4f41f4c375f5abb487dbd184fbc150f31994d762 && git merge --no-ff 6e9b2bee4ca544a1cc2569092bbe36c60efc754b
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.}

EagleBearerYS · 2026-11-03T15:14:18Z

Hi, a note on how this PR and #9742 fit together. Both clear native multisample depth or HTILE aliases a multisampled colour target whose samples are only on the host at the depth-plane index of the import, for an R32F colour target over the depth plane.

This PR takes the depth value from an R32F colour target whose newest clear or fill, in [0, 1], covers exactly the depth plane. It does so without HTILE, on the uniform-plane path, which fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742 doesn't touch, and with HTILE when every word proves that value. CheckCurrent re-proves the value.
fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742 (Ghost of Tsushima Director's Cut (PPSA02225), Dave the Diver (PPSA19115)) accepts one active colour target with the target's sample count over the depth plane, whatever it holds, when every HTILE word proves the target's own DB_DEPTH_CLEAR. Depth then reads DB_DEPTH_CLEAR. CheckCurrent keeps it while that target is unchanged (sameDepthAlias).

Where they meet: the index-0 branch of prepareDepthSampling and the index-0 keep in CheckCurrent, both in DepthSurface.cpp, and the test-list cmake, where the lists take the union. Git merges the two sets of test blocks in NativeMultisampleDepth.cpp on its own.

The HTILE case, a proposed composition. It's for #9742's merge of main after this PR lands; nothing changes here.

At import, both candidates are recorded: this PR's aliasDepth and fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742's depthAlias. The HTILE words are checked against aliasDepth first, and if they all prove it, the import reads it, as here. Otherwise, if fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742 recorded a sole alias and every word proves target.clearDepth, the import reads DB_DEPTH_CLEAR, as on fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742. Otherwise the import refuses with the existing HTILE line.
In CheckCurrent, fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742's sameDepthAlias comes first and keeps an alias that fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742 recorded. This PR's value re-proof then keeps the rest.
Why this order: this PR's proof is the more specific one, because it knows the plane's value and has HTILE prove it, so it decides whenever it applies. fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742's rule only adds the states this PR declines: a colour target whose value differs from what HTILE proves, or one that holds no known R32F clear.
One step of CheckHtileAliasedDepth changes under it: the fill of 1.0 with HTILE words proving 0.0, the target's own clear, is accepted and reads 0.0 instead of being refused. fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742's merge would turn that step into an accepted case and check the sampled value, since HTILE's clear decides what depth reads. If you see a reason that state should keep refusing, say so, and fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742 will keep the refusal.

Order: this PR first, then #9742, which takes the composition above when it merges main. #9742 also lands after #13021, which merges cleanly with this PR, so #13021 and this PR can go in either order. #15446's overlap with #9742 is the test list only.

edubox · 2026-11-03T16:47:06Z

The composition matches what the code can prove. On the CheckHtileAliasedDepth step with a 1.0 colour clear and HTILE words proving 0.0: I agree that #9742 can flip it to accepted and read DB_DEPTH_CLEAR once the broader rule lands. There is no reason to keep that state refused just for this PR, so #9742 is free to change the step and check the sampled 0.0. Landing this one first and then #9742 is fine by me.

A native multisample depth plane whose range exactly aliases a multisampled R32F colour target with a known clear or fill now takes that value instead of refusing the import. The HTILE path requires the metadata to prove the same value.

A depth alias no longer marks the separate stencil plane as aliased, so a stencil fill still needs its own tracked store and value check. The tests drop the drawn-target refusal, cover an HTILE that proves the alias value instead of the target clear, and check a stored and an unstored stencil fill next to the depth alias.

edubox · 2026-11-04T01:39:15Z

Rebased onto current origin/main (1c8c0efc8c) and pushed b6407b2431 to retrigger the hosted Windows job, whose only failure is the known infra case avplayer_http_patched_modules (not a changed file). No source change: the diff is the one at 6e9b2bee4c, only the base moved.

On this head: ANYPS5_REQUIRE_VULKAN=1 ANYPS5_NO_SHADER_CACHE=1 ctest --test-dir build -R '^agc_driver_native_(depth|multisample_depth)' passes 17/17 (including agc_driver_native_depth_uniform_plane_alias and agc_driver_native_depth_htile_plane_alias), and python3 tools/check_conventions.py --base origin/main reports all checks passed. The Vulkan validation comparison against main stands from the hosted AMD/NVIDIA runs on this code (no new validation IDs vs main).

TechLevelZero · 2026-11-04T02:19:45Z

✅ NVIDIA GPU test: passed

PR head b6407b2 merged onto main 1c8c0ef · tested 2026-11-04 02:19 UTC · 25 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	✅ 2,771 passed · 6 failed (0 new, 3 also fail on main, 3 unstable on main) · 50 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	2,827 of 3,842 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 (946 tests run on the GPU)
Slow tests (>30 s, GPU tests timed with validation layers)	⚠️ `agc_driver_pipeline_journal` 120.1 s

Unstable on main (3, these also failed in recent main runs on this machine; not counted against this PR)

agc_driver_pipeline_journal: failed in 4 of the last 10 main runs
agc_submit_progress: failed in 5 of the last 10 main runs
agc_submit_progress_no_timeline: failed in 4 of the last 10 main runs

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

agc_driver_image_table
guest_mmap_executable
guest_socket_options_multicast

Validation IDs already present on main (8 IDs in 533 tests)

VUID-VkDeviceCreateInfo-pNext-pNext (533 tests): agc_driver, agc_driver_address_passes, agc_driver_alias_copy_label, agc_driver_alpha_to_coverage, agc_driver_array_layer_target, agc_driver_bc_block_copy, agc_driver_bda_anchor_reads, agc_driver_bda_span_reads, and 525 more
VUID-vkCmdCopyImage-srcImage-01548 (7 tests): agc_driver_depth_clear_hle, agc_driver_depth_plane_buffer_memory, agc_driver_depth_plane_partial_store, agc_driver_depth_plane_sampling, agc_driver_depth_storage_dispatch, agc_driver_depth_surface_from_storage, agc_driver_dispatch_move_depth
VUID-vkCmdCopyImage-srcImage-01551 (7 tests): agc_driver_depth_clear_hle, agc_driver_depth_plane_buffer_memory, agc_driver_depth_plane_partial_store, agc_driver_depth_plane_sampling, agc_driver_depth_storage_dispatch, agc_driver_depth_surface_from_storage, agc_driver_dispatch_move_depth
VUID-VkPipelineColorBlendStateCreateInfo-pAttachments-00605 (1 test): agc_driver_srgb8_color_target
VUID-VkSamplerCreateInfo-addressModeU-01079 (1 test): agc_driver_image_gather_volume
VUID-VkSamplerCreateInfo-addressModeU-parameter (1 test): agc_driver_image_gather_volume
VUID-VkSamplerCreateInfo-addressModeV-parameter (1 test): agc_driver_image_gather_volume
VUID-VkSamplerCreateInfo-addressModeW-parameter (1 test): agc_driver_image_gather_volume

Skipped (50)

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_gcn_display_present: skipped, no display: Vulkan support is either not configured in SDL or not available in current SDL video driver (offscreen) or platform
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_scattered_runs: skipped, host imports are unavailable: /dev/udmabuf is not accessible
agc_driver_sparse_dma_buf_import: skipped, host imports are unavailable: /dev/udmabuf is not accessible
agc_driver_stencil_ref_export: skipped, the device lacks VK_EXT_shader_stencil_export
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 platform
vide

Reproduce

git fetch origin b6407b2431c74ee9e02cd9b80a0e0a076652815f 1c8c0efc8c41314402e6966da5bef6ea3d639d10
git checkout 1c8c0efc8c41314402e6966da5bef6ea3d639d10 && git merge --no-ff b6407b2431c74ee9e02cd9b80a0e0a076652815f
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.}

edubox · 2026-11-04T03:58:18Z

Verified the current head b6407b2431 locally (Linux x86-64, GCC 16.2.1, Release, RTX 4080 SUPER): all 16 agc_driver_native_depth tests pass with ANYPS5_REQUIRE_VULKAN=1, ANYPS5_NO_SHADER_CACHE=1 and VK_INSTANCE_LAYERS=VK_LAYER_KHRONOS_validation, including agc_driver_native_depth_uniform_plane_alias and agc_driver_native_depth_htile_plane_alias; no VUID- lines and no layer-missing message in the logs. tools/check_conventions.py --base origin/main is clean.

The retriggered hosted windows job failed again on avplayer_http_patched_modules, the same way as on the previous head: the build step never produced avplayer_http_patched_modules_tests.exe while the test was still selected, so file(COPY_FILE) at core/libs/tests/AvPlayerHttpModules.cmake:155 fails with No such file or directory (input). That looks deterministic for merges that select the test without building it, so re-pushing will not clear it; it needs an upstream fix to the test selection or the copy. Everything on this PR's own diff is green.

AI-assisted: yes

boykopovar#9742 has merged ahead of this branch. DepthSurface.cpp conflicts where both read a colour target over the depth plane at index 0, at import and in CheckCurrent, and the test list conflicts where both add registrations. The merge composes the two rules: - At import both are recorded, this branch's aliasDepth and boykopovar#9742's depthAlias. HTILE is checked against aliasDepth first. When it doesn't prove it and boykopovar#9742 recorded a sole alias, aliasDepth is dropped and every word must prove DB_DEPTH_CLEAR, as on boykopovar#9742. - CheckCurrent keeps boykopovar#9742's unchanged alias first, then this branch's value re-proof, which now compares bits, so a whole fill of -0.0 over a 0.0 clear still refuses, as boykopovar#9742's test expects. - In CheckHtileAliasedDepth, the fill of 1.0 over HTILE that proves 0.0 now imports and reads 0.0 under boykopovar#9742's rule instead of refusing. - Both set_tests_properties lines take both sides' names. - NativeMultisampleDepth.cpp keeps one depthAliasRefusal, boykopovar#9742's, which this branch defined again with the same text.

EagleBearerYS · 2026-11-04T07:42:21Z

Hi, a follow-up to my note above. #9742 merged first (aa186a7), so the order there is now the other way round. This PR conflicts with main in DepthSurface.cpp and in the native multisample depth test list, where the note said the two meet.

In case it saves you time, here's a merge of main 5b1d381 into this branch's b6407b2. It composes the two PRs the way the note proposed: 0103d305c2, on the prm/15443-main-5b1d-2 branch of EagleBearerYS/AnyPS5. Take it, redo it or leave it, whichever suits you.

What it does

At import, both candidates are recorded: this PR's aliasDepth and fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742's depthAlias.
- The HTILE words are checked against aliasDepth first. If every word proves it, the import reads it, as here.
- If they don't, and fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742 recorded a sole alias, aliasDepth is dropped, and every word must prove DB_DEPTH_CLEAR, as on fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742.
- Otherwise the import refuses with the existing HTILE line.
In CheckCurrent, fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742's sameDepthAlias keep comes first, then this PR's value re-proof.
One change to this PR's re-proof: it compares the bits, not the floats.
- fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742's NegativeZeroFill case refuses a whole fill of -0.0 over a colour target cleared to 0.0.
- With HTILE proving 0.0, the target's own value, this PR's path decides that import. A float comparison would then keep the -0.0 fill, since -0.0 == 0.0. Comparing the bits keeps fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742's refusal.
- If you'd rather accept that fill, fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742's test case is what would change.
In CheckHtileAliasedDepth, the step that fills 1.0 over HTILE proving 0.0 now imports and reads 0.0 under fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742's rule, as the note expected. So that step samples 0.0 instead of refusing, and the test's summary line says so.
The test list takes both sides' names.
The test file keeps one depthAliasRefusal, fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742's, which this branch also defines with the same text.

Tested on GitHub's runners, 10 runs (run).

The set: agc_driver_sample_multisample_target and every test matching ^agc_driver_native_depth or ^agc_driver_native_multisample. That takes in this PR's agc_driver_native_depth_uniform_plane_alias and agc_driver_native_depth_htile_plane_alias, and fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742's agc_driver_native_depth_aliased_depth_*.
Linux: 21 run, 20 pass, 1 skip (agc_driver_native_depth_bda_clear), 0 fail.
Windows: 21 run, 0 pass, 21 skip (no Vulkan device), 0 fail.
The libc and libkernel suites ran 302 and 326 tests on Linux and 332 and 344 on Windows, with 0 failed. They ran with APS5_SKIP_TIMING_TESTS=1, which relaxes precise_wait's median bounds on GitHub's hosted runners.

…rm-depth

edubox · 2026-11-04T09:50:26Z

Took the proposed composition from EagleBearerYS's 0103d305c2 as the base (it resolves DepthSurface.cpp and the native-depth test list against main 5b1d381d5a) and merged current origin/main (b11c7bf84f) on top, then pushed ae87f8ef1e.

The merge keeps both alias candidates (aliasDepth and the sole-target depthAlias), checks HTILE against aliasDepth first and falls back to DB_DEPTH_CLEAR, keeps sameDepthAlias in CheckCurrent before the value re-proof, compares the re-proof bits instead of floats, and takes the union of the test names.

Verification on an RTX 4080 SUPER with ANYPS5_REQUIRE_VULKAN=1, ANYPS5_NO_SHADER_CACHE=1 and VK_LAYER_KHRONOS_validation:

change: ctest --test-dir build -R '^agc_driver_native_depth' 19/19 pass;
main b11c7bf84f: the same selection 17/17 pass;
no VUID- lines on either side and the layer was inserted (no was not found but was requested).

python3 tools/check_conventions.py --base origin/main and git diff --check origin/main...HEAD are clean.

EagleBearerYS · 2026-11-04T09:54:52Z

Thanks for taking it, and for the RTX 4080 SUPER run with the validation layer on both sides. That settles the overlap with #9742 from our side.

TechLevelZero · 2026-11-04T14:52:21Z

✅ NVIDIA GPU test: passed

PR head ae87f8e merged onto main b11c7bf · tested 2026-11-04 14:52 UTC · 26 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	✅ 2,831 passed · 7 failed (0 new, 3 also fail on main, 4 unstable on main) · 49 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	2,887 of 3,894 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 (976 tests run on the GPU)
Slow tests (>30 s, GPU tests timed with validation layers)	⚠️ `agc_driver_pipeline_journal` 120.1 s

Unstable on main (4, these also failed in recent main runs on this machine; not counted against this PR)

agc_driver_pipeline_journal: failed in 3 of the last 10 main runs
agc_driver_submit_progress: failed in 3 of the last 10 main runs
agc_submit_progress: failed in 4 of the last 10 main runs
agc_submit_progress_no_timeline: failed in 3 of the last 10 main runs

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

agc_driver_image_table
guest_mmap_executable
guest_socket_options_multicast

Validation IDs already present on main (13 IDs in 548 tests)

VUID-VkDeviceCreateInfo-pNext-pNext (548 tests): agc_driver, agc_driver_address_draw_space_reuse, agc_driver_address_passes, agc_driver_alias_copy_label, agc_driver_alpha_to_coverage, agc_driver_array_layer_target, agc_driver_bc_block_copy, agc_driver_bda_anchor_reads, and 540 more
VUID-VkShaderModuleCreateInfo-pCode-08737 (8 tests): agc_driver_dpp16_operands, agc_driver_dpp8_pk_fmac, agc_driver_dx10_clamp, agc_driver_float16_fma_rounding, agc_driver_float16_unused_nan_payloads, agc_driver_packed_alu, agc_driver_packed_f16_denormal_modes, agc_driver_packed_float16_native
VUID-VkShaderModuleCreateInfo-pCode-08739 (8 tests): agc_driver_dpp16_operands, agc_driver_dpp8_pk_fmac, agc_driver_dx10_clamp, agc_driver_float16_fma_rounding, agc_driver_float16_unused_nan_payloads, agc_driver_packed_alu, agc_driver_packed_f16_denormal_modes, agc_driver_packed_float16_native
VUID-VkShaderModuleCreateInfo-pCode-08741 (8 tests): agc_driver_dpp16_operands, agc_driver_dpp8_pk_fmac, agc_driver_dx10_clamp, agc_driver_float16_fma_rounding, agc_driver_float16_unused_nan_payloads, agc_driver_packed_alu, agc_driver_packed_f16_denormal_modes, agc_driver_packed_float16_native
VUID-vkCmdCopyImage-srcImage-01548 (7 tests): agc_driver_depth_clear_hle, agc_driver_depth_plane_buffer_memory, agc_driver_depth_plane_partial_store, agc_driver_depth_plane_sampling, agc_driver_depth_storage_dispatch, agc_driver_depth_surface_from_storage, agc_driver_dispatch_move_depth
VUID-vkCmdCopyImage-srcImage-01551 (7 tests): agc_driver_depth_clear_hle, agc_driver_depth_plane_buffer_memory, agc_driver_depth_plane_partial_store, agc_driver_depth_plane_sampling, agc_driver_depth_storage_dispatch, agc_driver_depth_surface_from_storage, agc_driver_dispatch_move_depth
VUID-VkPipelineColorBlendStateCreateInfo-pAttachments-00605 (1 test): agc_driver_srgb8_color_target
VUID-VkPipelineShaderStageCreateInfo-module-parameter (1 test): agc_driver_float16_unused_nan_payloads
VUID-VkPipelineShaderStageCreateInfo-pSpecializationInfo-06849 (1 test): agc_driver_float16_unused_nan_payloads
VUID-VkSamplerCreateInfo-addressModeU-01079 (1 test): agc_driver_image_gather_volume
VUID-VkSamplerCreateInfo-addressModeU-parameter (1 test): agc_driver_image_gather_volume
VUID-VkSamplerCreateInfo-addressModeV-parameter (1 test): agc_driver_image_gather_volume
VUID-VkSamplerCreateInfo-addressModeW-parameter (1 test): agc_driver_image_gather_volume

Skipped (49)

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_gcn_display_present: skipped, no display: Vulkan support is either not configured in SDL or not available in current SDL video driver (offscreen) or platform
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_scattered_runs: skipped, host imports are unavailable: /dev/udmabuf is not accessible
agc_driver_sparse_dma_buf_import: skipped, host imports are unavailable: /dev/udmabuf is not accessible
agc_driver_stencil_ref_export: skipped, the device lacks VK_EXT_shader_stencil_export
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 platform
video_out_flip_eop_aftergpu: skipped, no display or Vulkan device: DisplayWindow: Vulkan support is either not configured in SDL or not av

Reproduce

git fetch origin ae87f8ef1ea6a3c44554cbb3c66fd185dbd8c072 b11c7bf84f5d2f636ff60166cc3371bd071e24d5
git checkout b11c7bf84f5d2f636ff60166cc3371bd071e24d5 && git merge --no-ff ae87f8ef1ea6a3c44554cbb3c66fd185dbd8c072
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.}

edubox · 2026-11-04T22:19:21Z

Re-checked the current head ae87f8ef1e against the review:

the stencil alias is tracked separately from the depth alias in CheckCurrent;
the new cases no longer require a throw for a colour target that has drawn;
the HTILE case with a target clear value different from the alias value is covered, and the sampled value is checked.

Focused checks on ae87f8ef1e, with ANYPS5_REQUIRE_VULKAN=1, ANYPS5_NO_SHADER_CACHE=1 and VK_INSTANCE_LAYERS=VK_LAYER_KHRONOS_validation:

agc_driver_native_depth_uniform_plane_alias: passed.
agc_driver_native_depth_htile_plane_alias: passed.
python3 tools/check_conventions.py --base origin/main: all checks passed.

No new change was pushed. The windows job shows cancelled after about 45 minutes, which is the runner limit rather than a test failure.

DonRyuzaki · 2026-11-04T23:55:08Z

Independent Windows / Radeon RX 7800 XT focused check of #15443 head ae87f8ef1ea6a3c44554cbb3c66fd185dbd8c072. I merged that head locally onto upstream main a992b803767e7411181c5c75f8e34bdfdd7aa110 (test merge aa61d4a58b71e0af750d7c45c6e2299436ff00fc) and built agc_driver_native_multisample_depth_tests successfully.

On the selected AMD Radeon RX 7800 XT (1002:747e, subgroup 64), both new CTests passed 3/3 runs each: agc_driver_native_depth_uniform_plane_alias and agc_driver_native_depth_htile_plane_alias. Each run checked 2×, 4× and 8× samples (checked=3 unavailable=0), with no CTest skip. The existing agc_driver_native_depth_uniform_planes and agc_driver_native_depth_stencil controls each passed once. The uniform alias fixture sampled the R32F clear at 0 and 1 and refused an out-of-range clear. The HTILE fixture sampled a matching alias, read DB_DEPTH_CLEAR when the colour clear differed from the HTILE proof, accepted a separately stored stencil fill and refused the unstored one.

I used ANYPS5_GPU=7800 XT, ANYPS5_REQUIRE_VULKAN=1, ANYPS5_NO_SHADER_CACHE=1 and requested VK_LAYER_KHRONOS_validation. The logs record the selected Radeon on every run. The validation callback emitted VUID-VkDeviceCreateInfo-pNext-pNext at device creation for unknown structure types 1000579000 and 1000574000 in the new tests and existing controls; no other VUID ID appears in these focused logs. I did not run a paired main baseline here, so this does not establish a validation delta.

This adds Windows RX 7800 XT hardware evidence to the Linux AMD and NVIDIA results already posted. No game or performance test was run. AI-assisted local test; build and verbose CTest logs are retained locally.

edubox · 2026-11-05T07:10:09Z

Re-checked ae87f8ef1e after the latest independent run. No code change is needed: the stencil alias is tracked separately from the depth alias, the new cases no longer pin the drawn-colour-target refusal, and the HTILE case with a different target clear is covered and samples the alias value.

On this machine (RTX 4080 SUPER, ANYPS5_REQUIRE_VULKAN=1, ANYPS5_NO_SHADER_CACHE=1, VK_LAYER_KHRONOS_validation inserted):

agc_driver_native_depth_uniform_plane_alias: passed.
agc_driver_native_depth_htile_plane_alias: passed.
No VUID- messages in either log.
python3 tools/check_conventions.py --base origin/main: passed.

Nothing was pushed. The windows job is still the runner's 45-minute cancellation, not a test failure.

AI-assisted: yes.

TechLevelZero · 2026-11-05T07:46:38Z

✅ AMD GPU test: passed

PR head ae87f8e merged onto main 4639eff · tested 2026-11-05 07:46 UTC · 20 min


GPU	AMD Radeon RX 6300 (RADV NAVI24), driver Mesa 25.2.8-0ubuntu0.24.04.4, Vulkan 1.4.318
Host	Ubuntu 26.04.1 LTS, kernel 7.0.0-42-generic, container ubuntu:24.04
Logs	build and test logs for this run
Build	✅ passed (Release, `-DANYPS5_ENABLE_SPIRV_TOOLS=ON`)
Tests	✅ 2,854 passed · 9 failed (0 new, 8 also fail on main, 1 unstable on main) · 67 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	2,930 of 3,962 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 (989 tests run on the GPU)
Slow tests (>30 s, GPU tests timed with validation layers)	⚠️ `agc_driver_pipeline_journal` 120.1 s, ⚠️ `public_headers` 64.5 s, ⚠️ `agc_driver_image_sample_unnormalized` 50.0 s, ⚠️ `agc_driver_lazy_resolved_abi` 36.9 s, ⚠️ `agc_driver_depth_clear_hle` 33.0 s

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

agc_driver_pipeline_journal: failed in 9 of the last 10 main runs

Failing on main too (8, not counted against this PR)

agc_driver_image_table
agc_driver_native_depth_bda_clear
agc_driver_native_multisample_depth
agc_driver_submit_progress
agc_submit_progress
agc_submit_progress_no_timeline
guest_mmap_executable
guest_socket_options_multicast

Validation IDs already present on main (4 IDs in 558 tests)

VUID-VkDeviceCreateInfo-pNext-pNext (558 tests): agc_driver, agc_driver_address_draw_space_reuse, agc_driver_address_passes, agc_driver_alias_copy_label, agc_driver_alpha_to_coverage, agc_driver_array_layer_target, agc_driver_bc_block_copy, agc_driver_bda_anchor_reads, and 550 more
VUID-vkCmdCopyImage-srcImage-01548 (4 tests): agc_driver_depth_clear_hle, agc_driver_depth_storage_dispatch, agc_driver_depth_surface_from_storage, agc_driver_dispatch_move_depth
VUID-vkCmdCopyImage-srcImage-01551 (4 tests): agc_driver_depth_clear_hle, agc_driver_depth_storage_dispatch, agc_driver_depth_surface_from_storage, agc_driver_dispatch_move_depth
VUID-VkPipelineColorBlendStateCreateInfo-pAttachments-00605 (2 tests): agc_driver_prepared_exports, agc_driver_srgb8_color_target

Skipped (67)

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_copied_writer_label_order: skipped, the device reports a minStorageBufferOffsetAlignment of 4, so the dispatch binds its record at +0x1004 in place and nothing is written back; the test needs an alignment above 4
agc_driver_float16_unused_nan_payloads: skipped, the device has no native f16 conversions with RTE rounding and kept f16 denormals
agc_driver_gcn_display_present: skipped, no display: Vulkan support is either not configured in SDL or not available in current SDL video driver (offscreen) or platform
agc_driver_image_table_vouched_lookups: skipped, the device reports 4 of 4 host-imported pages written after a GPU read, so a watched imported table cannot be vouched for
agc_driver_linear_target_direct_write_completion: skipped, the device does not keep host imports write-watched, so no completion stamp over the target is tracked
agc_driver_ngg_vertex_wave_reduction: skipped, s3 describes a 32-lane wave only on 32-lane subgroups, and the device has 64 lanes
agc_driver_recorded_draw_staging: skipped, the device does not keep host imports write-watched, so the target is never resident
agc_driver_scattered_runs: skipped, host imports are unavailable: /dev/udmabuf is not accessible
agc_driver_sparse_dma_buf_import: skipped, host imports are unavailable: /dev/udmabuf is not accessible
agc_driver_unit_shadow: skipped, the device does not import watched host memory for unit shadows
agc_driver_unit_shadow_budget: skipped, the device does not import watched host memory for unit shadows
agc_host_import_sparse: skipped, host imports are unavailable: /dev/udmabuf is not accessible
agc_idle_publish_wait: skipped, the device reports 4 of 4 host-imported pages written after a GPU read, so a forced watch of the import reads its submissions as CPU stores
agc_shadowed_key_stores: skipped, the device imports no watched host memory under a unit shadow
agc_small_tiled_shadow_label: skipped, host imports are unavailable: the test block is not imported and watched
agc_storage_clear_hand_over: skipped, the device imports no watched host memory for the clear-key hand-over
agc_storage_pending_key_refresh: skipped, refreshes over a pending key store are not testable here
agc_unit_shadow: skipped, the device does not import watched host memory for unit shadows
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_net_ctl_disconnected: 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_host_resolution_disabled: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_host_resolution_four_k: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_host_resolution_height_only: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_host_resolution_limit: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_host_resolution_malformed: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not a

Reproduce

git fetch origin ae87f8ef1ea6a3c44554cbb3c66fd185dbd8c072 4639eff7f4c23325a916875a0e37f1e312fa18f4
git checkout 4639eff7f4c23325a916875a0e37f1e312fa18f4 && git merge --no-ff ae87f8ef1ea6a3c44554cbb3c66fd185dbd8c072
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 12 --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 16 --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 AMD machine; it does not replace upstream CI. A new comment is posted for each new push. Full logs.}

…rm-depth

edubox · 2026-11-05T11:09:31Z

Updated the branch to current main (aafd88b74b) to pick up main's CI fix that builds the owner executable of wrapper tests, and to retrigger the hosted Windows job, which had been cancelled by the runner's 45-minute limit during the build. No PR source change; the composition merge is unchanged.

Focused checks on the merged head (80534241b6), RTX 4080 SUPER, ANYPS5_REQUIRE_VULKAN=1, ANYPS5_NO_SHADER_CACHE=1, VK_LAYER_KHRONOS_validation inserted:

20/20 agc_driver_native_(depth|multisample_depth) tests pass, including agc_driver_native_depth_uniform_plane_alias and agc_driver_native_depth_htile_plane_alias (3 sample-count cases each, none unavailable).
No VUID- lines and no was not found but was requested by env var VK_INSTANCE_LAYERS.
python3 tools/check_conventions.py --base origin/main and git diff --check origin/main...HEAD are clean.

AI-assisted: yes.

TechLevelZero · 2026-11-05T20:53:28Z

✅ NVIDIA GPU test: passed

PR head 8053424 merged onto main 080dbd0 · tested 2026-11-05 20:53 UTC · 31 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	✅ 2,943 passed · 3 failed (0 new, 3 also fail on main) · 54 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	3,000 of 4,077 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 (974 tests run on the GPU)
Slow tests (>30 s, GPU tests timed with validation layers)	⚠️ `agc_driver_pipeline_journal` 115.0 s

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

agc_driver_image_table
guest_mmap_executable
guest_socket_options_multicast

Validated but not compared (1, not in the main baseline's validation pass)

guest_kqueue_cleared_race

Validation IDs already present on main (9 IDs in 590 tests)

VUID-VkDeviceCreateInfo-pNext-pNext (590 tests): agc_compute_pipeline_replay, agc_driver, agc_driver_address_draw_space_reuse, agc_driver_address_passes, agc_driver_alias_copy_label, agc_driver_alpha_to_coverage, agc_driver_array_layer_target, agc_driver_bc_block_copy, and 582 more
VUID-VkShaderModuleCreateInfo-pCode-08737 (14 tests): agc_driver_dpp16_operands, agc_driver_dpp8_pk_fmac, agc_driver_dx10_clamp, agc_driver_f16_fma_denormal_modes, agc_driver_f16_fma_invalid_nan, agc_driver_float16_fma_rounding, agc_driver_float16_modifiers, agc_driver_float16_native, and 6 more
VUID-VkShaderModuleCreateInfo-pCode-08739 (14 tests): agc_driver_dpp16_operands, agc_driver_dpp8_pk_fmac, agc_driver_dx10_clamp, agc_driver_f16_fma_denormal_modes, agc_driver_f16_fma_invalid_nan, agc_driver_float16_fma_rounding, agc_driver_float16_modifiers, agc_driver_float16_native, and 6 more
VUID-VkShaderModuleCreateInfo-pCode-08741 (14 tests): agc_driver_dpp16_operands, agc_driver_dpp8_pk_fmac, agc_driver_dx10_clamp, agc_driver_f16_fma_denormal_modes, agc_driver_f16_fma_invalid_nan, agc_driver_float16_fma_rounding, agc_driver_float16_modifiers, agc_driver_float16_native, and 6 more
VUID-vkCmdCopyImage-srcImage-01548 (7 tests): agc_driver_depth_clear_hle, agc_driver_depth_plane_buffer_memory, agc_driver_depth_plane_partial_store, agc_driver_depth_plane_sampling, agc_driver_depth_storage_dispatch, agc_driver_depth_surface_from_storage, agc_driver_dispatch_move_depth
VUID-vkCmdCopyImage-srcImage-01551 (7 tests): agc_driver_depth_clear_hle, agc_driver_depth_plane_buffer_memory, agc_driver_depth_plane_partial_store, agc_driver_depth_plane_sampling, agc_driver_depth_storage_dispatch, agc_driver_depth_surface_from_storage, agc_driver_dispatch_move_depth
VUID-VkPipelineColorBlendStateCreateInfo-pAttachments-00605 (1 test): agc_driver_prepared_exports
VUID-VkPipelineShaderStageCreateInfo-module-parameter (1 test): agc_driver_float16_unused_nan_payloads
VUID-VkPipelineShaderStageCreateInfo-pSpecializationInfo-06849 (1 test): agc_driver_float16_unused_nan_payloads

Skipped (54)

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_gcn_display_present: skipped, no display: Vulkan support is either not configured in SDL or not available in current SDL video driver (offscreen) or platform
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_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_net_ctl_disconnected: 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_device_capability: 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_host_resolution_disabled: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_host_resolution_four_k: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_host_resolution_height_only: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_host_resolution_limit: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_host_resolution_malformed: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_host_resolution_width_only: 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

Reproduce

git fetch origin 80534241b6214dde51b3a16a22e77522c5427063 080dbd08c6f83be5a1768af7d3342d701bf34d21
git checkout 080dbd08c6f83be5a1768af7d3342d701bf34d21 && git merge --no-ff 80534241b6214dde51b3a16a22e77522c5427063
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.}

edubox · 2026-11-05T21:17:34Z

Status at 80534241b62:

The depth and stencil alias proofs are separate: aliasDepth and aliasStencil are tracked independently, and the stencil-fill branch still requires its own store/value proof.
The new cases no longer pin the drawn-colour-target refusal; the valid refusal for an out-of-range clear remains.
The differing-clear HTILE case is covered and samples DB_DEPTH_CLEAR when the HTILE proves a value different from the target's own clear.
The PR description includes the Vulkan validation comparison.

Verified locally without the Khronos layer (not installed on this host), with ANYPS5_REQUIRE_VULKAN=1 and ANYPS5_NO_SHADER_CACHE=1:

python3 tools/check_conventions.py --base origin/main: passed.
agc_driver_native_depth_separate_stencil, _aliased_stencil, _uniform_planes, _uniform_plane_alias, _htile_plane_alias: 5/5 passed.

github-actions Bot added build agc libSceAgc* labels Nov 3, 2026

edubox mentioned this pull request Nov 3, 2026

fix(agc): accept a stored refill of a pending HTILE clear #15446

Open

12 tasks done

EagleBearerYS mentioned this pull request Nov 3, 2026

fix(agc): accept a colour alias over a native multisample depth plane when HTILE proves its clear #9742

Merged

12 tasks done

edubox added 2 commits November 4, 2026 02:36

edubox force-pushed the fix/agc-aliased-uniform-depth branch from 6e9b2be to b6407b2 Compare November 4, 2026 01:38

github-actions Bot added the needs rebase Conflicts with main, rebase needed label Nov 4, 2026

Merge remote-tracking branch 'origin/main' into fix/agc-aliased-unifo…

ae87f8e

…rm-depth

github-actions Bot removed the needs rebase Conflicts with main, rebase needed label Nov 4, 2026

EagleBearerYS mentioned this pull request Nov 5, 2026

fix(agc): publish a reused native multisample depth target with the guest's undecided bytes, then retire its owner #15941

Draft

12 tasks done

Merge remote-tracking branch 'origin/main' into fix/agc-aliased-unifo…

8053424

…rm-depth

oneandonlydean self-assigned this Nov 6, 2026

Conversation

edubox commented Nov 3, 2026

What

Tested

Checklist

Uh oh!

oneandonlydean commented Nov 3, 2026

Uh oh!

edubox commented Nov 3, 2026

Uh oh!

RNK123 commented Nov 3, 2026

Uh oh!

TechLevelZero commented Nov 3, 2026

✅ AMD GPU test: passed

Uh oh!

TechLevelZero commented Nov 3, 2026

✅ NVIDIA GPU test: passed

Uh oh!

EagleBearerYS commented Nov 3, 2026

Uh oh!

edubox commented Nov 3, 2026

Uh oh!

edubox commented Nov 4, 2026

Uh oh!

TechLevelZero commented Nov 4, 2026

✅ NVIDIA GPU test: passed

Uh oh!

edubox commented Nov 4, 2026

Uh oh!

EagleBearerYS commented Nov 4, 2026

Uh oh!

edubox commented Nov 4, 2026

Uh oh!

EagleBearerYS commented Nov 4, 2026

Uh oh!

TechLevelZero commented Nov 4, 2026

✅ NVIDIA GPU test: passed

Uh oh!

edubox commented Nov 4, 2026

Uh oh!

DonRyuzaki commented Nov 4, 2026

Uh oh!

edubox commented Nov 5, 2026

Uh oh!

TechLevelZero commented Nov 5, 2026

✅ AMD GPU test: passed

Uh oh!

edubox commented Nov 5, 2026

Uh oh!

TechLevelZero commented Nov 5, 2026

✅ NVIDIA GPU test: passed

Uh oh!

edubox commented Nov 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

edubox commented Nov 5, 2026 •

edited

Loading