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 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]>
Queue-0 DMA_DATA packets drained the draw pipeline unless APS5_PIPELINE_DMA was set, so the packet worker waited on the committer at every small immediate store. Kena PPSA01746 issues 17-23k DMA_DATA packets per 10 s and 11329 of its 20134 pipeline drains per 10 s are opcode 0x50. enqueueDmaPacket now queues DMA_DATA stores and fills unconditionally, and keepsDrawPipeline keeps the pipeline for a copy or fill whose source and destination miss every queued write. The APS5_PIPELINE_DMA switch and its documentation are removed and the tests run the default path.
rewatchSettledImports passed an import's 64 KiB-rounded span to
Rewatch. For an import whose size is not a block multiple the span's
last block holds pages past the allocation, the walk fails on them, and
the import stayed compared for good. Pass the import's own pages; the
shared-block and live storage image checks still use the block span.
The second half of a pipelined draw (stage materialization, cache insert,
enqueue) runs on a finisher thread, overlapping the next draw's packet
work; draws whose stages could write memory are held back. Commits skip
work whose inputs did not change since the last commit, color metadata
passes are ordered through the pipeline, resource and storage-image proofs
are reused within one collect epoch until a write breaks them, and a shape
entry is refreshed in place at the second sight of the same read user words.
APS5_VERIFY_DRAW_PIPELINE checks reused proofs against a full check.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Under APS5_LAZY_SHADER_PREPARE=1 registration prepares nothing, but
ResolveShaderAbi still planned and prepared every compute or pixel stage
whose ABI the title resolved, so a title that resolves every shader's
ABI prepared them all at boot whether they ever drew.
With lazy preparation ResolveShaderAbi now plans the stage with the
ABI's state (its register and entry-point checks, as lazy registration
keeps them), records the ABI key, and queues the ABI's context and
primitive registers on the shader. The shader's first draw or dispatch
plans and prepares the queued ABIs with that state before it looks for
its artifact, so a shader that never draws or dispatches stays
unprepared and one that does gets the entries eager resolution would
have made. A stage whose plan fails is deferred as before. Without the
switch nothing changes; graphics-stage ABIs are still prepared when
they are resolved.
Test: agc_driver_lazy_resolved_abi resolves the ABI of four lazily
registered shaders and draws two. None holds prepared entries after
resolution, the two that never drew still hold none, and a drawn one
draws the same bytes as with eager preparation in a child process.
The sampled lookup kept the whole array's layer count: a texture over an
array of more than maxImageArrayLayers layers read and snapshotted the
whole surface and then threw in the Texture constructor, on every lookup,
even when the descriptor viewed a few slices. A menu that samples a
16x16 array of up to 8065 layers through its image tables spent
~55 ms per frame reading those surfaces only to reject them.
cachedTexture now applies the window the storage path uses
(StorageArrayWindow, renamed ArrayLayerWindow since both paths use it):
a view whose slices fit the limit is served as an array of its own
starting at its first slice, and a view whose slices alone exceed the
limit throws before anything is read. A 2D image is one layer whatever
its last array field holds and is no longer refused by the storage path.
agc_driver_image_array_window_past_layer_limit samples and loads three
slices at 8100-8102 of an 8103-layer 1D array (before: "a texture array
with more layers than maxImageArrayLayers is not implemented") and checks
that a view of all its layers is refused with the window's error on a
device whose limit is lower.
AI-assisted: yes
A volume's depthOrLastArray is its depth, bounded by maxImageDimension3D,
not by maxImageArrayLayers: a 2049 to 4096 deep volume is a valid image
and threw since the window was introduced. Reported by Dextroid17.
AI-assisted: yes
A PS5 array may hold up to 8192 layers; Vulkan devices allow fewer (2048 on
NVIDIA). A storage view of such an array created its whole image, with more
layers than the device allows: vkCreateImage is then invalid
(VUID-VkImageCreateInfo-arrayLayers-02256) and a store into a layer the device
did not back faults the GPU. Uncharted: Legacy of Thieves Collection (PPSA05684) views a 16x16 RGBA8
array of up to 7809 layers in slices of a few hundred to 2048 in its
orphanage chapter, which lost the device (Xid 109, write invalid) a few minutes in.
StorageArrayWindow describes the viewed slices as an array of their own
starting at the first viewed slice when the whole array is past the limit;
a non-XOR swizzle does not depend on the slice index, so the slices detile
alike. Views whose slices alone exceed the limit, XOR swizzles that mix in
the slice, and any other image past the limit now throw instead of creating
an invalid image.
AI-assisted: yes
An indirect draw drained queue 0's draw pipeline on entry and recorded every
record on the worker, so titles that draw mostly through DRAW_INDIRECT,
DRAW_INDEX_INDIRECT and DRAW_INDEX_INDIRECT_MULTI got no overlap between the
worker's capture of one draw and the committer's recording of the previous.
The draw now drains only when its records or its count overlap a write still
queued on the pipeline. Records drawn inline drain first, so they stay in
order, and the last drawn record is queued like a direct draw, carrying its
write ranges. APS5_NO_PIPELINED_INDIRECT=1 restores the drain on every
indirect draw.
agc_driver_pipelined_indirect_draw holds the committer with a blocking item and
checks that two indirect draws queue behind it and draw what a direct draw
does, and that a draw whose record a queued label writes waits for it.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
NVIDIA's Windows driver reads a host import of guest memory far slower than a buffer it allocated.
Since draw vertex and index ranges are bound in place from host imports (#2253, measured on Linux), Astro
Bot's Crash Site on Windows 11 with an RTX 5070 Ti spends 164 ms of every frame in draws on the GPU, and
45 ms with the ranges copied as before; with the same build, 4.71 fps in place against 7.52 copied. An
RTX 3060 shows the same (draws 32.3 against 24.3 ms per frame), while AMD's Windows driver reads them in
place about as fast as a copy (RX 9070 XT: 69.8 against 73.1 ms), so copying there only costs CPU.
The device decides once (DrawInputsInPlaceUsable, Context::drawInputsInPlace): ranges are copied with
VK_DRIVER_ID_NVIDIA_PROPRIETARY on Windows and bound in place everywhere else. APS5_DRAW_INPUTS_IN_PLACE=1
binds them in place and APS5_NO_DRAW_INPUTS_IN_PLACE=1 copies them on any device, the second winning.
agc_draw_inputs_in_place_driver checks the decision per driver and host; agc_draw_inputs_copied checks
that an aligned imported range is copied and its read not noted when the context copies, whatever the
inherited environment.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_01GN8FJFTedayRXLHythXkZ4
Devices with DenormFlushToZero for f32 run the mixed denormal-flush programs with and
without the hardware mode at startup; when every result matches, add, sub, mul, fma,
floor and ceil in the default rounding mode drop their software flush.
The data-hit verification keeps this branch's pending-span query on its ShaderMemory and takes #5958's CaptureInvocation, which retries a capture with converted samples.
Keeps this branch's tagVariantRuns declaration next to main's dumpRequest suffix parameter in Driver.hpp, and reads #6232's pointer-chain test words from its table's snapshot, where this branch keeps them.
#7527's selected-record limit test reads a record's descriptor words through the
table snapshot's shared words list; this branch keeps a snapshot's words per column,
so the test reads them from its column instead.
A per-game relinker-patches.toml (or --patches) replaces verified bytes in
the input executable and bundled modules before conversion. Entries are
checked against the expected bytes and against every in-place writer,
the ELF header range and relocation targets. An <output-stem>.patches.toml
manifest records the applied entries, and --verify-patches checks an input
or a converted output. With no patch file the output is unchanged.
Converted guest modules never ran their DT_PREINIT_ARRAY entries: the
reader validated the array and the unused-NID filter kept its targets,
but no writer emitted or ran it. ld.so runs a shared object's preinit
array only for the main executable, so the Linux output runs the
entries from the DT_INIT stub before the module initializer; the macOS
and Windows outputs run them before the initializer in their startup
stubs, and the deferred Windows output exports a thunk that runs them
before the initializer for runtime-loaded modules. A module with a
preinit array and no initializer is still rejected, since the Linux
stub exists only with one.
ARCHITECTURE.md and the libSceAgcDriver debt entry now say that a
settled Windows import under 1 MiB returns to the write watch only when
no storage image overlaps it, and why.
Main's #7269 keeps the synchronous settle's buffer, memory and fence
per device in ImportSettle, so clearing one device's imports releases
only that device's objects. This branch's settle batches move there
too: each device's ImportSettle holds its batch target, memory and
batches with their fences, so switching devices no longer drops a
device's batch handles, and clearing a device waits for and destroys
only its own batches. The per-import settle state is renamed
pendingSettles to stay clear of main's per-device map, and a submitted
settle records the device whose batch it is in.
Main's #6402 stores to image tables: a table with a written entry takes
the fresh path, as it does on main, and so does one without a sampled
table binding, whose entries the fresh path rejects. The shared place
step uses the binding as a pointer. Main's #6614 frees descriptor sets
from shared pools one by one, while this branch keeps only dedicated
pools, so both stay. #7101's bulk key read counts its unmapped-page
skips as unmapped, and its key words sit next to this branch's write
stamp and compared runs.
Main's #7101 reads a key buffer's words in one call where this branch
names the reason a narrowing is skipped. Keep main's bulk read and
count both of its unmapped-page skips as unmapped.
MayEvaluate and the evaluator track the first-lane reads they are inside, so a read reached again through a loop-carried phi or a constant slot is not evaluable instead of recursing.
Adrià