GFX11 renumbers the image formats past 29, so the raw codes 10, 37 and 1
read as other formats on the console's GFX10 numbering (37 is
11_11_10_UNORM there, 2_10_10_10_SNORM on gfx1100). Name them, and say
why the TYPE 9 depth compare control only answers at 32_FLOAT: gfx1100
compares 16_UNORM and 32_FLOAT only and answers zero for the others.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
The entry recorded TYPE 1 only from a gfx1010 measurement with format 22 and left other formats and
depth-compare reads open. Measured on a gfx1100 with the hardware oracle: image_sample_lz,
image_sample_b, image_load, image_sample_c_lz and image_get_resinfo return zero and no fault for
TYPE 0 and TYPE 1 over formats 22, 10, 37 and 1, with a nonzero base over readable memory and with a
zero base. A TYPE 9 control in the same kernel returns the stored texel, the raw dword, the declared
64 and 32, and 1.0 from image_sample_c_lz with an ALWAYS depth compare and format 22.
Co-Authored-By: Claude Opus 5 <[email protected]>
A delta read its blocks with ReadCommittedUnsynced while GPU work
accepted before the call could still write them, so a capture could
hold a block from before a DMA copy that the replay then wrote back over
the copy's result. A delta with blocks to read now waits up to 10 s for
the accepted submissions and records the serials after the wait; the
capture stops with a warning when the work does not finish. The Windows
quarantine of agc_driver_frame_replay is removed.
Fixes#10711
Co-Authored-By: Claude Opus 5.5 <[email protected]>
DecodeColorBuffer dropped the DCC address of a color target with more than
one mip, so the target's resident image had no keys: it was uploaded from
guest memory under the game's fast-clear keys, and no draw into it marked
them uncompressed. The game's textures over the same chain name the keys,
so they kept sampling the clear (0000 reads as (0,0,0,0)) whatever was
drawn. MXGP 26, Borderlands 4 and Forspoken render such targets.
- ColorTarget.chainDccAddress keeps the keys of a 2D single-layer chain;
dccAddress stays 0, so the per-draw single-mip key handling is unchanged.
- SurfaceForTarget gives the resident surface those keys, and the image
then treats them as a multi-mip storage image does: a uniform clear is
cleared on the GPU at upload, and the first write marks every key
uncompressed.
- A draw recipe's key check uses the surface's keys.
- sizeof(State) is 840: the recipe codec and CanonicalPipelineState do not
copy the field.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The graphics pipeline store kept a fixed 1024 entries and the pipeline journal replayed at most
1024 records, and a replayed pipeline that found the store full was dropped. A title whose
working set is larger rebuilt the rest at the draw on every run, and evicted pipelines it kept
using were rebuilt again and again.
The bound now starts at 1024 and doubles up to 4096: a journal replay that does not fit grows it
instead of dropping the pipeline, and so does eviction churn (a ghost list of evicted keys, checked
every bound/32 evictions: more than half of them came back). It never shrinks during a run;
APS5_PIPELINE_STORE_ENTRIES still fixes it. The journal loads its last 4096 records by default
(APS5_PIPELINE_JOURNAL_ENTRIES). Resident pipelines are bounded by count, not bytes.
agc_driver_recorder_tests --benchmark-pipeline-store-memory [count] (opt-in, not registered)
builds count graphics pipelines with one fragment module each through CachedPipeline and prints
process private bytes and VK_EXT_memory_budget heap usage before and after: with the triangle
shaders about 59 KiB host and no device memory per pipeline on an RTX 4090, 100 KiB host and
8 KiB device-local on a UHD 770.
Tests: agc_pipeline_store_growth (fails on main), agc_pipeline_store_fixed_bound,
agc_pipeline_store_replay_growth (replays on one thread: the mock Vulkan is not thread-safe).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_017SNUYRd7sRKZrMYpquAsnU
The Windows registration check fails when a test executable is built but no ctest test runs it. Put the executable under ANYPS5_RUN_QUARANTINED_FRAME_REPLAY as well, so OFF neither builds nor registers it and ON does both.
The recorder reads a delta block with ReadCommittedUnsynced while the second DMA copy into it is still in flight, so the capture can hold zeros and the replay then does not reproduce the second copy. The test failed intermittently on the Windows lane. Do not register it on Windows unless ANYPS5_RUN_QUARANTINED_FRAME_REPLAY is ON, and list it in the libSceAgcDriver technical debt.
TakeStorage copies a pending or live storage image over the whole depth
plane of a new single-sample surface, after the creation clear. On AMD's
Windows driver 32.0.31041.1004 (RX 7800 XT), one or two 8x8 blocks of a
D16 surface then tested as the clear value in about 1 of 10 runs of
agc_driver_depth_surface_from_storage, although a copy back of the image
read the copied values (#6630).
Without a stencil plane the copy now goes through an UNDEFINED to
TRANSFER_DST_OPTIMAL barrier on the depth aspect and back to GENERAL, so
the clear's state is discarded before the copy writes every texel. A
surface with a stencil plane keeps the previous path.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The journal was read whole into one buffer that the decoded modules
viewed, so a replay held the entire file (1 GiB at the default
APS5_PIPELINE_JOURNAL_MAX_MIB) until its last entry was built. The load
now parses the file a record at a time through positioned reads and keeps
it open; a replay thread reads an entry's modules again when it builds the
entry and rejects it when a record no longer reads back as loaded. The
replay report gains the peak module bytes held at once.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Adrià