perf(vk): a draw rebuilds only the descriptors whose sources changed (RX_VK_DESC_SPLIT) #1851

Closed
HumanBean wants to merge 2 commits from HumanBean/Longshot:perf/draw-cost into main
Owner

Summary

The render thread spends most of each frame translating the game's state into Vulkan, draw by draw. In Castle Town Market (~2,500 draws a frame) that is ~11 ms a frame on an RX 7900 XTX, and building and pushing each draw's descriptors is ~3.4 ms of it. This PR takes the first, safe step, on every GPU:

  1. Why a draw built its descriptors (RX_VK_PROF_FINE=1 only): new counters split the builds by cause. In the Market on an RX 580: another program or command buffer 0.53 a draw, a stage's textures moved 0.26, only the uniform buffers changed 0.14, storage buffers moved 0.02.
  2. A draw rebuilds only the descriptors whose sources changed (RX_VK_DESC_SPLIT, on; =0 the old way; =2 builds both ways and counts differences). With the same program and command buffer and only the uniform buffers changed, the uniform buffer writes are rebuilt in place and only they are pushed; textures and storage buffers are kept under strict conditions (handles from c2/c1 only, no tables, no texel buffers, no written images, unchanged generations and stamps). Intel Arc's push-every-uniform-buffer rule is kept.
    Bindless textures, which take on the 0.53 + 0.26 builds a draw that switch programs or textures, follow in their own pull request. Builds on dyst's #794 (kept descriptors per stage); see also #1763.

Read first: desc_split_keeps, build_uniform_buffers and the split in draw() (gpu/vk/vk_draw.cpp).

Testing

Tested: Linux, RX 580 (RADV), i7-9700K, 1080p: Kokiri Forest and Castle Town Market, standing and turning, with the split on and off; one Market session with RX_VK_DESC_SPLIT=2: hundreds of thousands of draws' kept textures and storage buffers compared with a whole build, 0 differed; no Vulkan errors. Renderer builds on Linux; Windows syntax check of the changed files with llvm-mingw.

Still to test:

  • Every platform: three areas against main, expecting equal or better fps and identical pictures (@Beanstalk)
  • Intel Arc: the Market, expecting no wrong textures with the split on (@Beanstalk)
  • Linux and Windows: mise run test, the new test_desc_split.py and the fixtures it touched, expecting them to pass (by hand: CI on this pull request)

Measurements

Verdict: correct (0 differences in the check) and even to slightly better on a GPU-bound RX 580, within noise; its gain is limited by design: 0.14 of ~0.95 builds a draw in the Market are uniform-buffer-only, an estimated 0.3-0.4 ms a frame where building descriptors is ~2.8 ms (RX 7900 XTX). RX_VK_DESC_SPLIT ships on: it is never slower here and its check found no difference. The diagnostics the diff names (RX_VK_TRACE_TEX, RX_VK_LOG_IMG, RX_VK_WATCH_PROG, RX_VK_DRAW_STATS, RX_VK_PEEK_PROG, RX_VK_LOG_TABLE, RX_VK_SLOW_DESC, RX_VK_ASYNC_IMPORT) are existing switches, unchanged and off by default; the split only turns itself off under them.

Baseline: this PR's own build with RX_VK_DESC_SPLIT=0, which builds every descriptor as main does, so the two runs differ only by the split; the Test Farm compares against main itself.

Setup: Linux, RX 580 (RADV), i7-9700K, 1080p; two interleaved runs each, this PR's build with RX_VK_DESC_SPLIT=0 (as main) against its default.

fps Before: main's behaviour, this PR @ de65a6b4 with RX_VK_DESC_SPLIT=0 After: this PR @ de65a6b4
Kokiri standing 50.4 50.5
Kokiri turning 40.0 41.0
Market standing 34.9 35.2
Market turning 53.5 53.4

Risks

A kept descriptor that should have been rebuilt would show a wrong texture or buffer: the keep rules are strict and RX_VK_DESC_SPLIT=2 checks them; RX_VK_DESC_SPLIT=0 turns it off. Kept storage buffers skip the bind's pool heat count, which may slightly change which pools get mirrors. Nothing changes for the player's files.

Checklist

Ready for review

  • One topic, on current main; the title is a Conventional Commit.
  • Tested on my own PC, in the game for a change the game runs, as Testing says.
  • mise run test passes (any failure that also fails on main is named).
  • Measured against main, or Measurements says why there is nothing to measure.
  • No game files, build output or hand-edited generated files; third_party/ untouched; nothing GPL-incompatible.

When it applies

  • A switch added or changed: it ships at the default the measurements chose, with its row in gpu/vk/switches.def.
  • A check under tests/ that reads changed source is updated, not weakened.
  • Host code builds on Windows (llvm-mingw) and Linux, with UTF-8 paths.
  • AI assistance recorded with Assisted-by: / Co-Authored-By: trailers.

Ready to merge

  • Everything in Still to test has run, its results in Tested.
  • Tier 1 hardware it can affect is tested (docs/dev/standards.md, "Hardware tiers").
  • Approved in review, every comment answered.
  • CI passes on the head; on current main, measured again or a comment says why the numbers still hold.

🤖 Generated with Claude Code

https://claude.ai/code/session_017vw8e3aEPBsd7QfUkaatjP

## Summary The render thread spends most of each frame translating the game's state into Vulkan, draw by draw. In Castle Town Market (~2,500 draws a frame) that is ~11 ms a frame on an RX 7900 XTX, and building and pushing each draw's descriptors is ~3.4 ms of it. This PR takes the first, safe step, on every GPU: 1. **Why a draw built its descriptors** (`RX_VK_PROF_FINE=1` only): new counters split the builds by cause. In the Market on an RX 580: another program or command buffer 0.53 a draw, a stage's textures moved 0.26, only the uniform buffers changed 0.14, storage buffers moved 0.02. 2. **A draw rebuilds only the descriptors whose sources changed** (`RX_VK_DESC_SPLIT`, on; `=0` the old way; `=2` builds both ways and counts differences). With the same program and command buffer and only the uniform buffers changed, the uniform buffer writes are rebuilt in place and only they are pushed; textures and storage buffers are kept under strict conditions (handles from c2/c1 only, no tables, no texel buffers, no written images, unchanged generations and stamps). Intel Arc's push-every-uniform-buffer rule is kept. Bindless textures, which take on the 0.53 + 0.26 builds a draw that switch programs or textures, follow in their own pull request. Builds on dyst's #794 (kept descriptors per stage); see also #1763. Read first: `desc_split_keeps`, `build_uniform_buffers` and the split in `draw()` (`gpu/vk/vk_draw.cpp`). ## Testing **Tested:** Linux, RX 580 (RADV), i7-9700K, 1080p: Kokiri Forest and Castle Town Market, standing and turning, with the split on and off; one Market session with `RX_VK_DESC_SPLIT=2`: hundreds of thousands of draws' kept textures and storage buffers compared with a whole build, 0 differed; no Vulkan errors. Renderer builds on Linux; Windows syntax check of the changed files with llvm-mingw. **Still to test:** - Every platform: three areas against main, expecting equal or better fps and identical pictures (@Beanstalk) - Intel Arc: the Market, expecting no wrong textures with the split on (@Beanstalk) - Linux and Windows: `mise run test`, the new `test_desc_split.py` and the fixtures it touched, expecting them to pass (by hand: CI on this pull request) ## Measurements **Verdict:** correct (0 differences in the check) and even to slightly better on a GPU-bound RX 580, within noise; its gain is limited by design: 0.14 of ~0.95 builds a draw in the Market are uniform-buffer-only, an estimated 0.3-0.4 ms a frame where building descriptors is ~2.8 ms (RX 7900 XTX). `RX_VK_DESC_SPLIT` ships on: it is never slower here and its check found no difference. The diagnostics the diff names (`RX_VK_TRACE_TEX`, `RX_VK_LOG_IMG`, `RX_VK_WATCH_PROG`, `RX_VK_DRAW_STATS`, `RX_VK_PEEK_PROG`, `RX_VK_LOG_TABLE`, `RX_VK_SLOW_DESC`, `RX_VK_ASYNC_IMPORT`) are existing switches, unchanged and off by default; the split only turns itself off under them. Baseline: this PR's own build with `RX_VK_DESC_SPLIT=0`, which builds every descriptor as main does, so the two runs differ only by the split; the Test Farm compares against main itself. Setup: Linux, RX 580 (RADV), i7-9700K, 1080p; two interleaved runs each, this PR's build with `RX_VK_DESC_SPLIT=0` (as main) against its default. | fps | Before: main's behaviour, this PR @ de65a6b4 with `RX_VK_DESC_SPLIT=0` | After: this PR @ de65a6b4 | |---|---|---| | Kokiri standing | 50.4 | 50.5 | | Kokiri turning | 40.0 | 41.0 | | Market standing | 34.9 | 35.2 | | Market turning | 53.5 | 53.4 | ## Risks A kept descriptor that should have been rebuilt would show a wrong texture or buffer: the keep rules are strict and `RX_VK_DESC_SPLIT=2` checks them; `RX_VK_DESC_SPLIT=0` turns it off. Kept storage buffers skip the bind's pool heat count, which may slightly change which pools get mirrors. Nothing changes for the player's files. ## Checklist ### Ready for review - [x] One topic, on current `main`; the title is a Conventional Commit. - [x] Tested on my own PC, in the game for a change the game runs, as Testing says. - [ ] `mise run test` passes (any failure that also fails on `main` is named). - [x] Measured against `main`, or Measurements says why there is nothing to measure. - [x] No game files, build output or hand-edited generated files; `third_party/` untouched; nothing GPL-incompatible. ### When it applies - [x] A switch added or changed: it ships at the default the measurements chose, with its row in `gpu/vk/switches.def`. - [x] A check under `tests/` that reads changed source is updated, not weakened. - [x] Host code builds on Windows (llvm-mingw) and Linux, with UTF-8 paths. - [x] AI assistance recorded with `Assisted-by:` / `Co-Authored-By:` trailers. ### Ready to merge - [ ] Everything in Still to test has run, its results in Tested. - [ ] Tier 1 hardware it can affect is tested (docs/dev/standards.md, "Hardware tiers"). - [ ] Approved in review, every comment answered. - [ ] CI passes on the head; on current `main`, measured again or a comment says why the numbers still hold. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_017vw8e3aEPBsd7QfUkaatjP
Under RX_VK_PROF_FINE, the fine lines now count, for each draw that builds its
descriptors, the first reason that holds: another program or command buffer;
reuse off, an early readback or a fallback after a dispatch; a stage's
tex_gen moved; a stage's ssbo_gen moved; registrations, an eviction or the
samplers moved; or the uniform buffers alone. They add up to the builds.

Also counted: the bank calls and fills of c2 (c0 is the rest), the steps of
the thread's uniform update generation between draws, where each texture
handle came from (c2, c1, a storage pointer's 0, or elsewhere: c3+ or a handle
program), and the draws whose textures fit the bindless plan's push words. A
third line gives how often the texture destroy, registrations and sampler
registrations generations move a frame. Uniform ranges copied with an
update's overlay were counted already.

DescBound keeps each stage's tex_gen and ssbo_gen as bound, for the reasons.

Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_017vw8e3aEPBsd7QfUkaatjP
perf(vk): a draw rebuilds only the descriptors whose sources changed (RX_VK_DESC_SPLIT)
Some checks failed
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been skipped
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been skipped
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been skipped
Third-party parts / As their notes say (pull_request) Successful in 14s
Third-party parts / The same as upstream (pull_request) Has been skipped
CI / Conventional Commit title (pull_request) Successful in 8s
Guide screenshots / Guide images and local UI source review (pull_request) Failing after 12s
CI / Workflow lint (actionlint) (pull_request) Successful in 13s
Website / Website on Cloudflare (pull_request_target) Has been skipped
PR description / Description follows the template (pull_request_target) Failing after 4s
beanstalk/windows+nvidia 1/1 fail
beanstalk/windows+intel 0/1 pass, 0 running, 2 queued
beanstalk/linux+amd 2/1 fail
beanstalk/linux+nvidia 2/1 fail
beanstalk/approval approved by HumanBean (reviewed)
beanstalk/windows+amd 2/1 fail
beanstalk/linux+intel 0/1 pass, 0 running, 2 queued
beanstalk/steamdeck 2/1 fail
beanstalk/amd+igpu 1/1 fail
CI / Website build and checks (pull_request) Has been cancelled
de65a6b467
Every object's own uniform buffer bind moves the bindings' generation, so a
draw of the same program built and compared all its descriptors again: its
textures, samplers and storage buffers as well as the uniform buffers that
moved. Now a draw of the program the draw before built for, in the same
command buffer, with the thread's writes still the ones it bound, builds only
its uniform buffers, in place at the positions the whole build gave them, and
bind_descriptors compares and pushes only those writes. Intel Arc's rule
stands: every uniform buffer write is still pushed there.

The textures and storage buffers are kept only when desc_split_keeps says a
whole build would make the same ones now:
- every stage is a graphics one (0 to 4), and no stage's tex_gen or ssbo_gen
  moved (its c2 and c0 banks are made from those bindings alone);
- textures: every handle word was read from c2 or c1, or is a storage
  pointer's word (no handle program or table, no c0 or c3+); no texel buffer
  view and no written storage image; no texture went (destroy generation), no
  sampler registration moved, the slots it resolved hold what they did (or,
  with Descriptor reuse off, no registration at all), the same texture epoch
  and the same frame of the samplers' book;
- storage buffers: all read, from c0 or c1, none the miss buffer, none at a
  virtual pool's address, all bound whole this time; and neither the read
  bindings' stamp (gpu_bind_stamp) nor the game pools' table
  (guest_pools_stamp, new) moved.
Anything else, or any doubt, builds whole. The sampled list a kept draw hands
the hazards is made again from every texture the build sampled, filtered by
gpu_written at its use.

Off for a fallback, with RX_VK_DESC_REUSE=never, for a program that writes a
storage buffer while an early readback waits, while a dump is armed, and when
RX_VK_TRACE_TEX, LOG_IMG, WATCH_PROG, DRAW_STATS, PEEK_PROG, LOG_TABLE,
SLOW_DESC or ASYNC_IMPORT is set. After DLAA's commands everything is pushed
and the next draw builds whole.

RX_VK_DESC_SPLIT=0: every descriptor built, as before. =2 (a test): built
whole, and what would have been kept compared with it, the differences counted
in the log and the first eight named. The fine profile counts the builds of
the uniform buffers alone.

tests/test_desc_split.py runs the keep rule, handle_banks and the sampled
list's refilter cut from vk_draw.cpp, and checks the draw's conditions;
sampler_admission_test.cpp's writer gets the kinds bind_descriptors reads.

Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_017vw8e3aEPBsd7QfUkaatjP
feat(vk): bindless building blocks: a slot allocator released on the GPU's timeline, and a SPIR-V rewrite to push-constant texture indices (not wired in)
Some checks failed
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been skipped
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been skipped
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been skipped
Third-party parts / The same as upstream (pull_request) Has been skipped
CI / Conventional Commit title (pull_request) Successful in 6s
PR description / Description follows the template (pull_request_target) Successful in 9s
Guide screenshots / Guide images and local UI source review (pull_request) Failing after 10s
CI / Workflow lint (actionlint) (pull_request) Successful in 10s
Third-party parts / As their notes say (pull_request) Successful in 11s
Website / Website on Cloudflare (pull_request_target) Has been skipped
CI / Website build and checks (pull_request) Has been cancelled
c10a1c2a6a
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_017vw8e3aEPBsd7QfUkaatjP
Author
Owner

Tested in game: Linux, RX 580 (RADV 25.x), i7-9700K, 1080p: Kokiri Forest and Castle Town Market standing and turning, split on and off, two runs each; one Market session with RX_VK_DESC_SPLIT=2 compared kept descriptors with whole builds over 100,000+ draws, 0 differed; no Vulkan errors.

Tested in game: Linux, RX 580 (RADV 25.x), i7-9700K, 1080p: Kokiri Forest and Castle Town Market standing and turning, split on and off, two runs each; one Market session with RX_VK_DESC_SPLIT=2 compared kept descriptors with whole builds over 100,000+ draws, 0 differed; no Vulkan errors.
HumanBean force-pushed perf/draw-cost from c10a1c2a6a
Some checks failed
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been skipped
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been skipped
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been skipped
Third-party parts / The same as upstream (pull_request) Has been skipped
CI / Conventional Commit title (pull_request) Successful in 6s
PR description / Description follows the template (pull_request_target) Successful in 9s
Guide screenshots / Guide images and local UI source review (pull_request) Failing after 10s
CI / Workflow lint (actionlint) (pull_request) Successful in 10s
Third-party parts / As their notes say (pull_request) Successful in 11s
Website / Website on Cloudflare (pull_request_target) Has been skipped
CI / Website build and checks (pull_request) Has been cancelled
to de65a6b467
Some checks failed
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been skipped
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been skipped
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been skipped
Third-party parts / As their notes say (pull_request) Successful in 14s
Third-party parts / The same as upstream (pull_request) Has been skipped
CI / Conventional Commit title (pull_request) Successful in 8s
Guide screenshots / Guide images and local UI source review (pull_request) Failing after 12s
CI / Workflow lint (actionlint) (pull_request) Successful in 13s
Website / Website on Cloudflare (pull_request_target) Has been skipped
PR description / Description follows the template (pull_request_target) Failing after 4s
beanstalk/windows+nvidia 1/1 fail
beanstalk/windows+intel 0/1 pass, 0 running, 2 queued
beanstalk/linux+amd 2/1 fail
beanstalk/linux+nvidia 2/1 fail
beanstalk/approval approved by HumanBean (reviewed)
beanstalk/windows+amd 2/1 fail
beanstalk/linux+intel 0/1 pass, 0 running, 2 queued
beanstalk/steamdeck 2/1 fail
beanstalk/amd+igpu 1/1 fail
CI / Website build and checks (pull_request) Has been cancelled
2026-11-20 07:18:22 +00:00
Compare
HumanBean changed title from WIP: perf(vk): cheaper draws: descriptors rebuilt only where their sources changed, then bindless textures to WIP: perf(vk): a draw rebuilds only the descriptors whose sources changed (RX_VK_DESC_SPLIT) 2026-11-20 07:18:43 +00:00
Author
Owner

@Beanstalk Please test this PR against main on every platform: windows+amd, windows+nvidia, windows+intel, linux+amd, linux+nvidia, linux+intel, the steamdeck, amd+igpu. Settings vsync=2 (Off) and frame_rate_cap=6 (Uncapped), Threaded rendering Auto. From the kokiri-start save: Kokiri Forest, Hyrule Field and Castle Town Market, in each 20 s standing then 15 s turning the camera, two game starts a side; make sure each side has reached the area before measuring it. Measure fps, p99 frame time, stutters and crashes, and compare pictures with main (the change must draw exactly as main does; Intel Arc especially, which has a rule about pushing uniform buffers this change keeps).

@Beanstalk Please test this PR against main on every platform: windows+amd, windows+nvidia, windows+intel, linux+amd, linux+nvidia, linux+intel, the steamdeck, amd+igpu. Settings vsync=2 (Off) and frame_rate_cap=6 (Uncapped), Threaded rendering Auto. From the `kokiri-start` save: Kokiri Forest, Hyrule Field and Castle Town Market, in each 20 s standing then 15 s turning the camera, two game starts a side; make sure each side has reached the area before measuring it. Measure fps, p99 frame time, stutters and crashes, and compare pictures with main (the change must draw exactly as main does; Intel Arc especially, which has a rule about pushing uniform buffers this change keeps).
Member
Beanstalk test farm · an earlier run: commit de65a6b (the newer results are in a comment below)

Commit de65a6b · tier reviewed · approved by HumanBean · dashboard

What will be tested · AI plan (gpt-6.1-terra)

Against main, two starts per side visit Kokiri Forest, Hyrule Field and Market from the Kokiri save, with V-Sync off and the frame rate uncapped. Each area has a 20-second standing measurement, a frame-counted camera turn and a picture comparison; turning frames contribute to run-wide hitch counts rather than a separate FPS measurement. Crashes, any picture difference, or regressions beyond the FPS, p99, stutter and hitch limits fail the plan. Each tester run takes roughly 25 minutes; the Windows and Linux Intel targets will wait for matching testers. Threaded rendering remains at its default because the farm exposes no setting for it. No tester online now matches windows+intel, linux+intel: those targets wait for one.

  • Where: windows+amd (1-2 reports), windows+nvidia (1-2 reports), windows+intel (1-2 reports), linux+amd (1-2 reports), linux+nvidia (1-2 reports), linux+intel (1-2 reports), steamdeck (1-2 reports), amd+igpu (1-2 reports).
  • three-areas: from the repository's save kokiri-start (settings vsync=2, frame_rate_cap=6), then teleport to Kokiri Forest, measure kokiri-standing for 20 s (after 5 s warm-up), take the picture kokiri to compare with main, move the stick, teleport to Hyrule Field, measure field-standing for 20 s (after 5 s warm-up), take the picture field to compare with main, teleport to Market, measure market-standing for 20 s (after 5 s warm-up), take the picture market to compare with main; 2 runs on each side.
  • Measured: crashes, frame times, stutters and hitches, draws and frame rate.
  • The PR fails if: it crashes; or frame rate is more than 10% below main; or a picture differs from main by more than 0%; or stutters is more than 20% above main; or hitches over 50 ms in the whole run is more than 20% above main; or frame time (p99) is more than 15% above main.
  • Time: about 25 min of game runs per tester (4 game starts), plus the two builds.

Started at 07:19 UTC.

âš  No tester has a PC for linux+intel (none seen in 14 days): it waits until one signs in.

Target Reports Crash Frame p99 (ms) vs main VRAM peak (MB) vs main Result
windows+amd · 1: windows+amd+amd-proprietary+dgpu AMD Radeon RX 9060 XT, 2.0.416; 2: windows+amd+amd-proprietary+dgpu AMD Radeon RX 9070 XT, 2.0.416 2/1 1 2 none 13.1 vs 12.1 (+8%) — fail ⚠ outlier
windows+nvidia · 1: windows+nvidia+nvidia-proprietary+dgpu NVIDIA GeForce RTX 3080 Ti, 621.24.0; 2: windows+nvidia+nvidia-proprietary+dgpu NVIDIA GeForce RTX 3050 Ti Laptop GPU, 621.24.0 2/1 1 2 none 11.3 vs 11.3 (+1%) — fail
windows+intel 0/1 — — — queued
linux+amd · 1: linux+amd+radv+igpu AMD Radeon 780M Graphics (RADV PHOENIX), 26.1.99; 2: linux+amd+radv+igpu+handheld AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99 2/1 1 2 none 34.3 vs 35.9 (−5%) — fail ⚠ outlier
linux+nvidia · 1: linux+nvidia+nvidia-proprietary+dgpu NVIDIA GeForce GTX 1070, 580.196.7; 2: linux+nvidia+nvidia-proprietary+dgpu NVIDIA GeForce RTX 5070, 620.70.6 2/1 1 2 none 27.7 vs 18.0 (+54%) — fail ⚠ outlier
linux+intel 0/1 — — — queued
steamdeck · 1: linux+amd+radv+igpu+handheld AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99; 2: linux+amd+radv+igpu+handheld AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99 2/1 1 2 none 45.2 vs 45.7 (−1%) — fail ⚠ outlier
amd+igpu · 1: linux+amd+radv+igpu AMD Radeon 780M Graphics (RADV PHOENIX), 26.1.99; 2: linux+amd+radv+igpu AMD Radeon 760M Graphics (RADV PHOENIX), 26.0.8 2/1 1 2 none 53.2 vs 52.7 (+1%) — fail ⚠ outlier

Measurements · windows+amd

Verdict: fail on windows+amd (2/1 report): three-areas field-standing.fps_mean.min_delta_pct (-32.6%); three-areas screenshot.max_diff; three-areas kokiri-standing.stutter_count.max_delta_pct (+300.0%) failed.

Hardware: 1: windows+amd+amd-proprietary+dgpu AMD Radeon RX 9060 XT, 2.0.416 AMD Ryzen 5 5600 6-Core Processor, 32 GB RAM; 2: windows+amd+amd-proprietary+dgpu AMD Radeon RX 9070 XT, 2.0.416 AMD Ryzen 7 5800X3D 8-Core Processor, 32 GB RAM
Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them
Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports
How to reproduce: the farm's plan above (@Beanstalk on this PR runs it again)

Before: main @ a3565ac After: this PR @ de65a6b
three-areas · field-standing · frame rate 104.9 89.3 (−15%)
three-areas · kokiri-standing · frame rate 98.2 98.0 (−0%)
three-areas · market-standing · frame rate 66.6 65.7 (−1%)
three-areas · field-standing · stutters 0.2 0.5 (+100%)
three-areas · kokiri-standing · stutters 3.2 2.5 (−23%)
three-areas · market-standing · stutters 0.0 1.0
three-areas · hitches over 50 ms in the whole run 19.0 20.2 (+7%)
three-areas · field-standing · frame time (p99) (ms) 12.1 13.1 (+8%)
three-areas · kokiri-standing · frame time (p99) (ms) 14.0 13.9 (−1%)
three-areas · market-standing · frame time (p99) (ms) 19.2 21.2 (+10%)

Flags and settings

None: the plan varied no setting, switch or build option.

Markdown for the PR description
## Measurements

**Verdict:** fail on windows+amd (2/1 report): `three-areas field-standing.fps_mean.min_delta_pct (-32.6%)`; `three-areas screenshot.max_diff`; `three-areas kokiri-standing.stutter_count.max_delta_pct (+300.0%)` failed.

Hardware: 1: windows+amd+amd-proprietary+dgpu `AMD Radeon RX 9060 XT, 2.0.416` `AMD Ryzen 5 5600 6-Core Processor, 32 GB RAM`; 2: windows+amd+amd-proprietary+dgpu `AMD Radeon RX 9070 XT, 2.0.416` `AMD Ryzen 7 5800X3D 8-Core Processor, 32 GB RAM`  
Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them  
Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports  
How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again)

| | Before: main @ a3565ac | After: this PR @ de65a6b |
|---|---|---|
| three-areas · field-standing · frame rate | 104.9 | 89.3 (−15%) |
| three-areas · kokiri-standing · frame rate | 98.2 | 98.0 (−0%) |
| three-areas · market-standing · frame rate | 66.6 | 65.7 (−1%) |
| three-areas · field-standing · stutters | 0.2 | 0.5 (+100%) |
| three-areas · kokiri-standing · stutters | 3.2 | 2.5 (−23%) |
| three-areas · market-standing · stutters | 0.0 | 1.0 |
| three-areas · hitches over 50 ms in the whole run | 19.0 | 20.2 (+7%) |
| three-areas · field-standing · frame time (p99) (ms) | 12.1 | 13.1 (+8%) |
| three-areas · kokiri-standing · frame time (p99) (ms) | 14.0 | 13.9 (−1%) |
| three-areas · market-standing · frame time (p99) (ms) | 19.2 | 21.2 (+10%) |

### Flags and settings

None: the plan varied no setting, switch or build option.

Measurements · windows+nvidia

Verdict: fail on windows+nvidia (2/1 report): three-areas market-standing.stutter_count.max_delta_pct failed.

Hardware: 1: windows+nvidia+nvidia-proprietary+dgpu NVIDIA GeForce RTX 3080 Ti, 621.24.0 AMD Ryzen 9 5900X 12-Core Processor, 64 GB RAM; 2: windows+nvidia+nvidia-proprietary+dgpu NVIDIA GeForce RTX 3050 Ti Laptop GPU, 621.24.0 AMD Ryzen 7 5800H with Radeon Graphics, 16 GB RAM
Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them
Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports
How to reproduce: the farm's plan above (@Beanstalk on this PR runs it again)

Before: main @ a3565ac After: this PR @ de65a6b
three-areas · field-standing · frame rate 107.4 107.1 (−0%)
three-areas · kokiri-standing · frame rate 106.5 106.5 (−0%)
three-areas · market-standing · frame rate 78.6 78.5 (−0%)
three-areas · field-standing · stutters 0.0 0.0
three-areas · kokiri-standing · stutters 0.0 0.0
three-areas · market-standing · stutters 0.0 1.0
three-areas · hitches over 50 ms in the whole run 15.5 14.5 (−6%)
three-areas · field-standing · frame time (p99) (ms) 11.3 11.3 (+1%)
three-areas · kokiri-standing · frame time (p99) (ms) 11.9 11.6 (−2%)
three-areas · market-standing · frame time (p99) (ms) 16.3 16.7 (+3%)

Flags and settings

None: the plan varied no setting, switch or build option.

Markdown for the PR description
## Measurements

**Verdict:** fail on windows+nvidia (2/1 report): `three-areas market-standing.stutter_count.max_delta_pct` failed.

Hardware: 1: windows+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce RTX 3080 Ti, 621.24.0` `AMD Ryzen 9 5900X 12-Core Processor, 64 GB RAM`; 2: windows+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce RTX 3050 Ti Laptop GPU, 621.24.0` `AMD Ryzen 7 5800H with Radeon Graphics, 16 GB RAM`  
Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them  
Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports  
How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again)

| | Before: main @ a3565ac | After: this PR @ de65a6b |
|---|---|---|
| three-areas · field-standing · frame rate | 107.4 | 107.1 (−0%) |
| three-areas · kokiri-standing · frame rate | 106.5 | 106.5 (−0%) |
| three-areas · market-standing · frame rate | 78.6 | 78.5 (−0%) |
| three-areas · field-standing · stutters | 0.0 | 0.0 |
| three-areas · kokiri-standing · stutters | 0.0 | 0.0 |
| three-areas · market-standing · stutters | 0.0 | 1.0 |
| three-areas · hitches over 50 ms in the whole run | 15.5 | 14.5 (−6%) |
| three-areas · field-standing · frame time (p99) (ms) | 11.3 | 11.3 (+1%) |
| three-areas · kokiri-standing · frame time (p99) (ms) | 11.9 | 11.6 (−2%) |
| three-areas · market-standing · frame time (p99) (ms) | 16.3 | 16.7 (+3%) |

### Flags and settings

None: the plan varied no setting, switch or build option.

Measurements · linux+amd

Verdict: fail on linux+amd (2/1 report): three-areas screenshot.max_diff failed.

Hardware: 1: linux+amd+radv+igpu AMD Radeon 780M Graphics (RADV PHOENIX), 26.1.99 AMD Ryzen 7 7840HS w/ Radeon 780M Graphics, 24 GB RAM; 2: linux+amd+radv+igpu+handheld AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99 AMD Custom APU 0405, 12 GB RAM
Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them
Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports
How to reproduce: the farm's plan above (@Beanstalk on this PR runs it again)

Before: main @ a3565ac After: this PR @ de65a6b
three-areas · field-standing · frame rate 35.6 35.6 (−0%)
three-areas · kokiri-standing · frame rate 37.6 38.0 (+1%)
three-areas · market-standing · frame rate 27.0 27.8 (+3%)
three-areas · field-standing · stutters 0.0 0.0
three-areas · kokiri-standing · stutters 0.2 0.0 (−100%)
three-areas · market-standing · stutters 8.8 0.5 (−94%)
three-areas · hitches over 50 ms in the whole run 563.8 646.2 (+15%)
three-areas · field-standing · frame time (p99) (ms) 35.9 34.3 (−5%)
three-areas · kokiri-standing · frame time (p99) (ms) 37.3 36.6 (−2%)
three-areas · market-standing · frame time (p99) (ms) 82.0 60.2 (−27%)

Flags and settings

None: the plan varied no setting, switch or build option.

Markdown for the PR description
## Measurements

**Verdict:** fail on linux+amd (2/1 report): `three-areas screenshot.max_diff` failed.

Hardware: 1: linux+amd+radv+igpu `AMD Radeon 780M Graphics (RADV PHOENIX), 26.1.99` `AMD Ryzen 7 7840HS w/ Radeon 780M Graphics, 24 GB RAM`; 2: linux+amd+radv+igpu+handheld `AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99` `AMD Custom APU 0405, 12 GB RAM`  
Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them  
Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports  
How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again)

| | Before: main @ a3565ac | After: this PR @ de65a6b |
|---|---|---|
| three-areas · field-standing · frame rate | 35.6 | 35.6 (−0%) |
| three-areas · kokiri-standing · frame rate | 37.6 | 38.0 (+1%) |
| three-areas · market-standing · frame rate | 27.0 | 27.8 (+3%) |
| three-areas · field-standing · stutters | 0.0 | 0.0 |
| three-areas · kokiri-standing · stutters | 0.2 | 0.0 (−100%) |
| three-areas · market-standing · stutters | 8.8 | 0.5 (−94%) |
| three-areas · hitches over 50 ms in the whole run | 563.8 | 646.2 (+15%) |
| three-areas · field-standing · frame time (p99) (ms) | 35.9 | 34.3 (−5%) |
| three-areas · kokiri-standing · frame time (p99) (ms) | 37.3 | 36.6 (−2%) |
| three-areas · market-standing · frame time (p99) (ms) | 82.0 | 60.2 (−27%) |

### Flags and settings

None: the plan varied no setting, switch or build option.

Measurements · linux+nvidia

Verdict: fail on linux+nvidia (2/1 report): three-areas screenshot.max_diff; three-areas kokiri-standing.stutter_count.max_delta_pct (+100.0%); three-areas kokiri-standing.fps_mean.min_delta_pct (-32.7%) failed.

Hardware: 1: linux+nvidia+nvidia-proprietary+dgpu NVIDIA GeForce GTX 1070, 580.196.7 Intel(R) Core(TM) i7-8750H CPU @ 2.20GHz, 32 GB RAM; 2: linux+nvidia+nvidia-proprietary+dgpu NVIDIA GeForce RTX 5070, 620.70.6 AMD Ryzen 5 9600X 6-Core Processor, 32 GB RAM
Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them
Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports
How to reproduce: the farm's plan above (@Beanstalk on this PR runs it again)

Before: main @ a3565ac After: this PR @ de65a6b
three-areas · field-standing · frame rate 78.2 62.8 (−20%)
three-areas · kokiri-standing · frame rate 77.1 60.6 (−21%)
three-areas · market-standing · frame rate 62.6 60.2 (−4%)
three-areas · field-standing · stutters 0.0 9.2
three-areas · kokiri-standing · stutters 0.5 1.2 (+150%)
three-areas · market-standing · stutters 0.5 0.5 (+0%)
three-areas · hitches over 50 ms in the whole run 24.8 127.0 (+413%)
three-areas · field-standing · frame time (p99) (ms) 18.0 27.7 (+54%)
three-areas · kokiri-standing · frame time (p99) (ms) 19.2 29.7 (+55%)
three-areas · market-standing · frame time (p99) (ms) 22.8 24.6 (+8%)

Flags and settings

None: the plan varied no setting, switch or build option.

Markdown for the PR description
## Measurements

**Verdict:** fail on linux+nvidia (2/1 report): `three-areas screenshot.max_diff`; `three-areas kokiri-standing.stutter_count.max_delta_pct (+100.0%)`; `three-areas kokiri-standing.fps_mean.min_delta_pct (-32.7%)` failed.

Hardware: 1: linux+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce GTX 1070, 580.196.7` `Intel(R) Core(TM) i7-8750H CPU @ 2.20GHz, 32 GB RAM`; 2: linux+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce RTX 5070, 620.70.6` `AMD Ryzen 5 9600X 6-Core Processor, 32 GB RAM`  
Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them  
Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports  
How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again)

| | Before: main @ a3565ac | After: this PR @ de65a6b |
|---|---|---|
| three-areas · field-standing · frame rate | 78.2 | 62.8 (−20%) |
| three-areas · kokiri-standing · frame rate | 77.1 | 60.6 (−21%) |
| three-areas · market-standing · frame rate | 62.6 | 60.2 (−4%) |
| three-areas · field-standing · stutters | 0.0 | 9.2 |
| three-areas · kokiri-standing · stutters | 0.5 | 1.2 (+150%) |
| three-areas · market-standing · stutters | 0.5 | 0.5 (+0%) |
| three-areas · hitches over 50 ms in the whole run | 24.8 | 127.0 (+413%) |
| three-areas · field-standing · frame time (p99) (ms) | 18.0 | 27.7 (+54%) |
| three-areas · kokiri-standing · frame time (p99) (ms) | 19.2 | 29.7 (+55%) |
| three-areas · market-standing · frame time (p99) (ms) | 22.8 | 24.6 (+8%) |

### Flags and settings

None: the plan varied no setting, switch or build option.

Measurements · steamdeck

Verdict: fail on steamdeck (2/1 report): three-areas screenshot.max_diff; three-areas market-standing.stutter_count.max_delta_pct (+200.0%); three-areas market-standing.frametime_p99_ms.max_delta_pct (+30.9%) failed.

Hardware: 1: linux+amd+radv+igpu+handheld AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99 AMD Custom APU 0405, 12 GB RAM; 2: linux+amd+radv+igpu+handheld AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99 AMD Custom APU 0405, 16 GB RAM
Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them
Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports
How to reproduce: the farm's plan above (@Beanstalk on this PR runs it again)

Before: main @ a3565ac After: this PR @ de65a6b
three-areas · field-standing · frame rate 23.9 23.8 (−0%)
three-areas · kokiri-standing · frame rate 23.5 24.5 (+4%)
three-areas · market-standing · frame rate 17.7 17.4 (−2%)
three-areas · field-standing · stutters 0.0 0.0
three-areas · kokiri-standing · stutters 3.2 0.2 (−92%)
three-areas · market-standing · stutters 2.8 2.5 (−9%)
three-areas · hitches over 50 ms in the whole run 1,337 1,439 (+8%)
three-areas · field-standing · frame time (p99) (ms) 45.7 45.2 (−1%)
three-areas · kokiri-standing · frame time (p99) (ms) 72.0 49.6 (−31%)
three-areas · market-standing · frame time (p99) (ms) 97.0 93.4 (−4%)

Flags and settings

None: the plan varied no setting, switch or build option.

Markdown for the PR description
## Measurements

**Verdict:** fail on steamdeck (2/1 report): `three-areas screenshot.max_diff`; `three-areas market-standing.stutter_count.max_delta_pct (+200.0%)`; `three-areas market-standing.frametime_p99_ms.max_delta_pct (+30.9%)` failed.

Hardware: 1: linux+amd+radv+igpu+handheld `AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99` `AMD Custom APU 0405, 12 GB RAM`; 2: linux+amd+radv+igpu+handheld `AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99` `AMD Custom APU 0405, 16 GB RAM`  
Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them  
Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports  
How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again)

| | Before: main @ a3565ac | After: this PR @ de65a6b |
|---|---|---|
| three-areas · field-standing · frame rate | 23.9 | 23.8 (−0%) |
| three-areas · kokiri-standing · frame rate | 23.5 | 24.5 (+4%) |
| three-areas · market-standing · frame rate | 17.7 | 17.4 (−2%) |
| three-areas · field-standing · stutters | 0.0 | 0.0 |
| three-areas · kokiri-standing · stutters | 3.2 | 0.2 (−92%) |
| three-areas · market-standing · stutters | 2.8 | 2.5 (−9%) |
| three-areas · hitches over 50 ms in the whole run | 1,337 | 1,439 (+8%) |
| three-areas · field-standing · frame time (p99) (ms) | 45.7 | 45.2 (−1%) |
| three-areas · kokiri-standing · frame time (p99) (ms) | 72.0 | 49.6 (−31%) |
| three-areas · market-standing · frame time (p99) (ms) | 97.0 | 93.4 (−4%) |

### Flags and settings

None: the plan varied no setting, switch or build option.

Measurements · amd+igpu

Verdict: fail on amd+igpu (2/1 report): three-areas screenshot.max_diff failed.

Hardware: 1: linux+amd+radv+igpu AMD Radeon 780M Graphics (RADV PHOENIX), 26.1.99 AMD Ryzen 7 7840HS w/ Radeon 780M Graphics, 24 GB RAM; 2: linux+amd+radv+igpu AMD Radeon 760M Graphics (RADV PHOENIX), 26.0.8 AMD Ryzen 5 230 w/ Radeon 760M Graphics, 16 GB RAM
Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them
Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports
How to reproduce: the farm's plan above (@Beanstalk on this PR runs it again)

Before: main @ a3565ac After: this PR @ de65a6b
three-areas · field-standing · frame rate 33.4 33.3 (−0%)
three-areas · kokiri-standing · frame rate 35.9 37.0 (+3%)
three-areas · market-standing · frame rate 27.4 27.7 (+1%)
three-areas · field-standing · stutters 0.0 0.0
three-areas · kokiri-standing · stutters 0.0 0.0
three-areas · market-standing · stutters 4.0 0.2 (−94%)
three-areas · hitches over 50 ms in the whole run 1,014 1,018 (+0%)
three-areas · field-standing · frame time (p99) (ms) 52.7 53.2 (+1%)
three-areas · kokiri-standing · frame time (p99) (ms) 46.1 36.7 (−20%)
three-areas · market-standing · frame time (p99) (ms) 63.9 59.3 (−7%)

Flags and settings

None: the plan varied no setting, switch or build option.

Markdown for the PR description
## Measurements

**Verdict:** fail on amd+igpu (2/1 report): `three-areas screenshot.max_diff` failed.

Hardware: 1: linux+amd+radv+igpu `AMD Radeon 780M Graphics (RADV PHOENIX), 26.1.99` `AMD Ryzen 7 7840HS w/ Radeon 780M Graphics, 24 GB RAM`; 2: linux+amd+radv+igpu `AMD Radeon 760M Graphics (RADV PHOENIX), 26.0.8` `AMD Ryzen 5 230 w/ Radeon 760M Graphics, 16 GB RAM`  
Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them  
Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports  
How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again)

| | Before: main @ a3565ac | After: this PR @ de65a6b |
|---|---|---|
| three-areas · field-standing · frame rate | 33.4 | 33.3 (−0%) |
| three-areas · kokiri-standing · frame rate | 35.9 | 37.0 (+3%) |
| three-areas · market-standing · frame rate | 27.4 | 27.7 (+1%) |
| three-areas · field-standing · stutters | 0.0 | 0.0 |
| three-areas · kokiri-standing · stutters | 0.0 | 0.0 |
| three-areas · market-standing · stutters | 4.0 | 0.2 (−94%) |
| three-areas · hitches over 50 ms in the whole run | 1,014 | 1,018 (+0%) |
| three-areas · field-standing · frame time (p99) (ms) | 52.7 | 53.2 (+1%) |
| three-areas · kokiri-standing · frame time (p99) (ms) | 46.1 | 36.7 (−20%) |
| three-areas · market-standing · frame time (p99) (ms) | 63.9 | 59.3 (−7%) |

### Flags and settings

None: the plan varied no setting, switch or build option.
<!-- beanstalk:previous --> <details><summary>Beanstalk test farm · an earlier run: commit <code>de65a6b</code> (the newer results are in a comment below)</summary> Commit `de65a6b` · tier **reviewed** · approved by HumanBean · [dashboard](https://beanstalk.magicbean.dev/prs/1851) **What will be tested** · AI plan (`gpt-6.1-terra`) Against main, two starts per side visit Kokiri Forest, Hyrule Field and Market from the Kokiri save, with V-Sync off and the frame rate uncapped. Each area has a 20-second standing measurement, a frame-counted camera turn and a picture comparison; turning frames contribute to run-wide hitch counts rather than a separate FPS measurement. Crashes, any picture difference, or regressions beyond the FPS, p99, stutter and hitch limits fail the plan. Each tester run takes roughly 25 minutes; the Windows and Linux Intel targets will wait for matching testers. Threaded rendering remains at its default because the farm exposes no setting for it. No tester online now matches windows+intel, linux+intel: those targets wait for one. - **Where:** windows+amd (1-2 reports), windows+nvidia (1-2 reports), windows+intel (1-2 reports), linux+amd (1-2 reports), linux+nvidia (1-2 reports), linux+intel (1-2 reports), steamdeck (1-2 reports), amd+igpu (1-2 reports). - **three-areas:** from the repository's save `kokiri-start` (settings vsync=2, frame_rate_cap=6), then teleport to Kokiri Forest, measure kokiri-standing for 20 s (after 5 s warm-up), take the picture kokiri to compare with main, move the stick, teleport to Hyrule Field, measure field-standing for 20 s (after 5 s warm-up), take the picture field to compare with main, teleport to Market, measure market-standing for 20 s (after 5 s warm-up), take the picture market to compare with main; 2 runs on each side. - **Measured:** crashes, frame times, stutters and hitches, draws and frame rate. - **The PR fails if:** it crashes; or frame rate is more than 10% below main; or a picture differs from main by more than 0%; or stutters is more than 20% above main; or hitches over 50 ms in the whole run is more than 20% above main; or frame time (p99) is more than 15% above main. - **Time:** about 25 min of game runs per tester (4 game starts), plus the two builds. Started at 07:19 UTC. ⚠ No tester has a PC for `linux+intel` (none seen in 14 days): it waits until one signs in. | Target | Reports | Crash | Frame p99 (ms) vs main | VRAM peak (MB) vs main | Result | | --- | --- | --- | --- | --- | --- | | windows+amd · 1: windows+amd+amd-proprietary+dgpu `AMD Radeon RX 9060 XT, 2.0.416`; 2: windows+amd+amd-proprietary+dgpu `AMD Radeon RX 9070 XT, 2.0.416` | 2/1 [1](https://beanstalk.magicbean.dev/reports/rp_ef03b35ed4c07db9) [2](https://beanstalk.magicbean.dev/reports/rp_32e0336adeef032b) | none | 13.1 vs 12.1 (+8%) | — | fail ⚠ outlier | | windows+nvidia · 1: windows+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce RTX 3080 Ti, 621.24.0`; 2: windows+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce RTX 3050 Ti Laptop GPU, 621.24.0` | 2/1 [1](https://beanstalk.magicbean.dev/reports/rp_f1069db6cb66a087) [2](https://beanstalk.magicbean.dev/reports/rp_535cf67ee54791a3) | none | 11.3 vs 11.3 (+1%) | — | fail | | windows+intel | 0/1 | — | — | — | queued | | linux+amd · 1: linux+amd+radv+igpu `AMD Radeon 780M Graphics (RADV PHOENIX), 26.1.99`; 2: linux+amd+radv+igpu+handheld `AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99` | 2/1 [1](https://beanstalk.magicbean.dev/reports/rp_c71a57918c742acb) [2](https://beanstalk.magicbean.dev/reports/rp_a3c261d0f654f2fe) | none | 34.3 vs 35.9 (−5%) | — | fail ⚠ outlier | | linux+nvidia · 1: linux+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce GTX 1070, 580.196.7`; 2: linux+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce RTX 5070, 620.70.6` | 2/1 [1](https://beanstalk.magicbean.dev/reports/rp_31f3e268b6731a24) [2](https://beanstalk.magicbean.dev/reports/rp_19bd9b369a20de1b) | none | 27.7 vs 18.0 (+54%) | — | fail ⚠ outlier | | linux+intel | 0/1 | — | — | — | queued | | steamdeck · 1: linux+amd+radv+igpu+handheld `AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99`; 2: linux+amd+radv+igpu+handheld `AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99` | 2/1 [1](https://beanstalk.magicbean.dev/reports/rp_ba785ed29919d611) [2](https://beanstalk.magicbean.dev/reports/rp_8a4a3ab2b1df212e) | none | 45.2 vs 45.7 (−1%) | — | fail ⚠ outlier | | amd+igpu · 1: linux+amd+radv+igpu `AMD Radeon 780M Graphics (RADV PHOENIX), 26.1.99`; 2: linux+amd+radv+igpu `AMD Radeon 760M Graphics (RADV PHOENIX), 26.0.8` | 2/1 [1](https://beanstalk.magicbean.dev/reports/rp_1f0b0dc5c38290f1) [2](https://beanstalk.magicbean.dev/reports/rp_185cda4c49990d46) | none | 53.2 vs 52.7 (+1%) | — | fail ⚠ outlier | #### Measurements · windows+amd **Verdict:** fail on windows+amd (2/1 report): `three-areas field-standing.fps_mean.min_delta_pct (-32.6%)`; `three-areas screenshot.max_diff`; `three-areas kokiri-standing.stutter_count.max_delta_pct (+300.0%)` failed. Hardware: 1: windows+amd+amd-proprietary+dgpu `AMD Radeon RX 9060 XT, 2.0.416` `AMD Ryzen 5 5600 6-Core Processor, 32 GB RAM`; 2: windows+amd+amd-proprietary+dgpu `AMD Radeon RX 9070 XT, 2.0.416` `AMD Ryzen 7 5800X3D 8-Core Processor, 32 GB RAM` Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again) | | Before: main @ a3565ac | After: this PR @ de65a6b | |---|---|---| | three-areas · field-standing · frame rate | 104.9 | 89.3 (−15%) | | three-areas · kokiri-standing · frame rate | 98.2 | 98.0 (−0%) | | three-areas · market-standing · frame rate | 66.6 | 65.7 (−1%) | | three-areas · field-standing · stutters | 0.2 | 0.5 (+100%) | | three-areas · kokiri-standing · stutters | 3.2 | 2.5 (−23%) | | three-areas · market-standing · stutters | 0.0 | 1.0 | | three-areas · hitches over 50 ms in the whole run | 19.0 | 20.2 (+7%) | | three-areas · field-standing · frame time (p99) (ms) | 12.1 | 13.1 (+8%) | | three-areas · kokiri-standing · frame time (p99) (ms) | 14.0 | 13.9 (−1%) | | three-areas · market-standing · frame time (p99) (ms) | 19.2 | 21.2 (+10%) | ### Flags and settings None: the plan varied no setting, switch or build option. <details><summary>Markdown for the PR description</summary> ```markdown ## Measurements **Verdict:** fail on windows+amd (2/1 report): `three-areas field-standing.fps_mean.min_delta_pct (-32.6%)`; `three-areas screenshot.max_diff`; `three-areas kokiri-standing.stutter_count.max_delta_pct (+300.0%)` failed. Hardware: 1: windows+amd+amd-proprietary+dgpu `AMD Radeon RX 9060 XT, 2.0.416` `AMD Ryzen 5 5600 6-Core Processor, 32 GB RAM`; 2: windows+amd+amd-proprietary+dgpu `AMD Radeon RX 9070 XT, 2.0.416` `AMD Ryzen 7 5800X3D 8-Core Processor, 32 GB RAM` Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again) | | Before: main @ a3565ac | After: this PR @ de65a6b | |---|---|---| | three-areas · field-standing · frame rate | 104.9 | 89.3 (−15%) | | three-areas · kokiri-standing · frame rate | 98.2 | 98.0 (−0%) | | three-areas · market-standing · frame rate | 66.6 | 65.7 (−1%) | | three-areas · field-standing · stutters | 0.2 | 0.5 (+100%) | | three-areas · kokiri-standing · stutters | 3.2 | 2.5 (−23%) | | three-areas · market-standing · stutters | 0.0 | 1.0 | | three-areas · hitches over 50 ms in the whole run | 19.0 | 20.2 (+7%) | | three-areas · field-standing · frame time (p99) (ms) | 12.1 | 13.1 (+8%) | | three-areas · kokiri-standing · frame time (p99) (ms) | 14.0 | 13.9 (−1%) | | three-areas · market-standing · frame time (p99) (ms) | 19.2 | 21.2 (+10%) | ### Flags and settings None: the plan varied no setting, switch or build option. ``` </details> #### Measurements · windows+nvidia **Verdict:** fail on windows+nvidia (2/1 report): `three-areas market-standing.stutter_count.max_delta_pct` failed. Hardware: 1: windows+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce RTX 3080 Ti, 621.24.0` `AMD Ryzen 9 5900X 12-Core Processor, 64 GB RAM`; 2: windows+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce RTX 3050 Ti Laptop GPU, 621.24.0` `AMD Ryzen 7 5800H with Radeon Graphics, 16 GB RAM` Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again) | | Before: main @ a3565ac | After: this PR @ de65a6b | |---|---|---| | three-areas · field-standing · frame rate | 107.4 | 107.1 (−0%) | | three-areas · kokiri-standing · frame rate | 106.5 | 106.5 (−0%) | | three-areas · market-standing · frame rate | 78.6 | 78.5 (−0%) | | three-areas · field-standing · stutters | 0.0 | 0.0 | | three-areas · kokiri-standing · stutters | 0.0 | 0.0 | | three-areas · market-standing · stutters | 0.0 | 1.0 | | three-areas · hitches over 50 ms in the whole run | 15.5 | 14.5 (−6%) | | three-areas · field-standing · frame time (p99) (ms) | 11.3 | 11.3 (+1%) | | three-areas · kokiri-standing · frame time (p99) (ms) | 11.9 | 11.6 (−2%) | | three-areas · market-standing · frame time (p99) (ms) | 16.3 | 16.7 (+3%) | ### Flags and settings None: the plan varied no setting, switch or build option. <details><summary>Markdown for the PR description</summary> ```markdown ## Measurements **Verdict:** fail on windows+nvidia (2/1 report): `three-areas market-standing.stutter_count.max_delta_pct` failed. Hardware: 1: windows+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce RTX 3080 Ti, 621.24.0` `AMD Ryzen 9 5900X 12-Core Processor, 64 GB RAM`; 2: windows+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce RTX 3050 Ti Laptop GPU, 621.24.0` `AMD Ryzen 7 5800H with Radeon Graphics, 16 GB RAM` Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again) | | Before: main @ a3565ac | After: this PR @ de65a6b | |---|---|---| | three-areas · field-standing · frame rate | 107.4 | 107.1 (−0%) | | three-areas · kokiri-standing · frame rate | 106.5 | 106.5 (−0%) | | three-areas · market-standing · frame rate | 78.6 | 78.5 (−0%) | | three-areas · field-standing · stutters | 0.0 | 0.0 | | three-areas · kokiri-standing · stutters | 0.0 | 0.0 | | three-areas · market-standing · stutters | 0.0 | 1.0 | | three-areas · hitches over 50 ms in the whole run | 15.5 | 14.5 (−6%) | | three-areas · field-standing · frame time (p99) (ms) | 11.3 | 11.3 (+1%) | | three-areas · kokiri-standing · frame time (p99) (ms) | 11.9 | 11.6 (−2%) | | three-areas · market-standing · frame time (p99) (ms) | 16.3 | 16.7 (+3%) | ### Flags and settings None: the plan varied no setting, switch or build option. ``` </details> #### Measurements · linux+amd **Verdict:** fail on linux+amd (2/1 report): `three-areas screenshot.max_diff` failed. Hardware: 1: linux+amd+radv+igpu `AMD Radeon 780M Graphics (RADV PHOENIX), 26.1.99` `AMD Ryzen 7 7840HS w/ Radeon 780M Graphics, 24 GB RAM`; 2: linux+amd+radv+igpu+handheld `AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99` `AMD Custom APU 0405, 12 GB RAM` Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again) | | Before: main @ a3565ac | After: this PR @ de65a6b | |---|---|---| | three-areas · field-standing · frame rate | 35.6 | 35.6 (−0%) | | three-areas · kokiri-standing · frame rate | 37.6 | 38.0 (+1%) | | three-areas · market-standing · frame rate | 27.0 | 27.8 (+3%) | | three-areas · field-standing · stutters | 0.0 | 0.0 | | three-areas · kokiri-standing · stutters | 0.2 | 0.0 (−100%) | | three-areas · market-standing · stutters | 8.8 | 0.5 (−94%) | | three-areas · hitches over 50 ms in the whole run | 563.8 | 646.2 (+15%) | | three-areas · field-standing · frame time (p99) (ms) | 35.9 | 34.3 (−5%) | | three-areas · kokiri-standing · frame time (p99) (ms) | 37.3 | 36.6 (−2%) | | three-areas · market-standing · frame time (p99) (ms) | 82.0 | 60.2 (−27%) | ### Flags and settings None: the plan varied no setting, switch or build option. <details><summary>Markdown for the PR description</summary> ```markdown ## Measurements **Verdict:** fail on linux+amd (2/1 report): `three-areas screenshot.max_diff` failed. Hardware: 1: linux+amd+radv+igpu `AMD Radeon 780M Graphics (RADV PHOENIX), 26.1.99` `AMD Ryzen 7 7840HS w/ Radeon 780M Graphics, 24 GB RAM`; 2: linux+amd+radv+igpu+handheld `AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99` `AMD Custom APU 0405, 12 GB RAM` Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again) | | Before: main @ a3565ac | After: this PR @ de65a6b | |---|---|---| | three-areas · field-standing · frame rate | 35.6 | 35.6 (−0%) | | three-areas · kokiri-standing · frame rate | 37.6 | 38.0 (+1%) | | three-areas · market-standing · frame rate | 27.0 | 27.8 (+3%) | | three-areas · field-standing · stutters | 0.0 | 0.0 | | three-areas · kokiri-standing · stutters | 0.2 | 0.0 (−100%) | | three-areas · market-standing · stutters | 8.8 | 0.5 (−94%) | | three-areas · hitches over 50 ms in the whole run | 563.8 | 646.2 (+15%) | | three-areas · field-standing · frame time (p99) (ms) | 35.9 | 34.3 (−5%) | | three-areas · kokiri-standing · frame time (p99) (ms) | 37.3 | 36.6 (−2%) | | three-areas · market-standing · frame time (p99) (ms) | 82.0 | 60.2 (−27%) | ### Flags and settings None: the plan varied no setting, switch or build option. ``` </details> #### Measurements · linux+nvidia **Verdict:** fail on linux+nvidia (2/1 report): `three-areas screenshot.max_diff`; `three-areas kokiri-standing.stutter_count.max_delta_pct (+100.0%)`; `three-areas kokiri-standing.fps_mean.min_delta_pct (-32.7%)` failed. Hardware: 1: linux+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce GTX 1070, 580.196.7` `Intel(R) Core(TM) i7-8750H CPU @ 2.20GHz, 32 GB RAM`; 2: linux+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce RTX 5070, 620.70.6` `AMD Ryzen 5 9600X 6-Core Processor, 32 GB RAM` Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again) | | Before: main @ a3565ac | After: this PR @ de65a6b | |---|---|---| | three-areas · field-standing · frame rate | 78.2 | 62.8 (−20%) | | three-areas · kokiri-standing · frame rate | 77.1 | 60.6 (−21%) | | three-areas · market-standing · frame rate | 62.6 | 60.2 (−4%) | | three-areas · field-standing · stutters | 0.0 | 9.2 | | three-areas · kokiri-standing · stutters | 0.5 | 1.2 (+150%) | | three-areas · market-standing · stutters | 0.5 | 0.5 (+0%) | | three-areas · hitches over 50 ms in the whole run | 24.8 | 127.0 (+413%) | | three-areas · field-standing · frame time (p99) (ms) | 18.0 | 27.7 (+54%) | | three-areas · kokiri-standing · frame time (p99) (ms) | 19.2 | 29.7 (+55%) | | three-areas · market-standing · frame time (p99) (ms) | 22.8 | 24.6 (+8%) | ### Flags and settings None: the plan varied no setting, switch or build option. <details><summary>Markdown for the PR description</summary> ```markdown ## Measurements **Verdict:** fail on linux+nvidia (2/1 report): `three-areas screenshot.max_diff`; `three-areas kokiri-standing.stutter_count.max_delta_pct (+100.0%)`; `three-areas kokiri-standing.fps_mean.min_delta_pct (-32.7%)` failed. Hardware: 1: linux+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce GTX 1070, 580.196.7` `Intel(R) Core(TM) i7-8750H CPU @ 2.20GHz, 32 GB RAM`; 2: linux+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce RTX 5070, 620.70.6` `AMD Ryzen 5 9600X 6-Core Processor, 32 GB RAM` Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again) | | Before: main @ a3565ac | After: this PR @ de65a6b | |---|---|---| | three-areas · field-standing · frame rate | 78.2 | 62.8 (−20%) | | three-areas · kokiri-standing · frame rate | 77.1 | 60.6 (−21%) | | three-areas · market-standing · frame rate | 62.6 | 60.2 (−4%) | | three-areas · field-standing · stutters | 0.0 | 9.2 | | three-areas · kokiri-standing · stutters | 0.5 | 1.2 (+150%) | | three-areas · market-standing · stutters | 0.5 | 0.5 (+0%) | | three-areas · hitches over 50 ms in the whole run | 24.8 | 127.0 (+413%) | | three-areas · field-standing · frame time (p99) (ms) | 18.0 | 27.7 (+54%) | | three-areas · kokiri-standing · frame time (p99) (ms) | 19.2 | 29.7 (+55%) | | three-areas · market-standing · frame time (p99) (ms) | 22.8 | 24.6 (+8%) | ### Flags and settings None: the plan varied no setting, switch or build option. ``` </details> #### Measurements · steamdeck **Verdict:** fail on steamdeck (2/1 report): `three-areas screenshot.max_diff`; `three-areas market-standing.stutter_count.max_delta_pct (+200.0%)`; `three-areas market-standing.frametime_p99_ms.max_delta_pct (+30.9%)` failed. Hardware: 1: linux+amd+radv+igpu+handheld `AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99` `AMD Custom APU 0405, 12 GB RAM`; 2: linux+amd+radv+igpu+handheld `AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99` `AMD Custom APU 0405, 16 GB RAM` Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again) | | Before: main @ a3565ac | After: this PR @ de65a6b | |---|---|---| | three-areas · field-standing · frame rate | 23.9 | 23.8 (−0%) | | three-areas · kokiri-standing · frame rate | 23.5 | 24.5 (+4%) | | three-areas · market-standing · frame rate | 17.7 | 17.4 (−2%) | | three-areas · field-standing · stutters | 0.0 | 0.0 | | three-areas · kokiri-standing · stutters | 3.2 | 0.2 (−92%) | | three-areas · market-standing · stutters | 2.8 | 2.5 (−9%) | | three-areas · hitches over 50 ms in the whole run | 1,337 | 1,439 (+8%) | | three-areas · field-standing · frame time (p99) (ms) | 45.7 | 45.2 (−1%) | | three-areas · kokiri-standing · frame time (p99) (ms) | 72.0 | 49.6 (−31%) | | three-areas · market-standing · frame time (p99) (ms) | 97.0 | 93.4 (−4%) | ### Flags and settings None: the plan varied no setting, switch or build option. <details><summary>Markdown for the PR description</summary> ```markdown ## Measurements **Verdict:** fail on steamdeck (2/1 report): `three-areas screenshot.max_diff`; `three-areas market-standing.stutter_count.max_delta_pct (+200.0%)`; `three-areas market-standing.frametime_p99_ms.max_delta_pct (+30.9%)` failed. Hardware: 1: linux+amd+radv+igpu+handheld `AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99` `AMD Custom APU 0405, 12 GB RAM`; 2: linux+amd+radv+igpu+handheld `AMD Custom GPU 0405 (RADV VANGOGH), 26.1.99` `AMD Custom APU 0405, 16 GB RAM` Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again) | | Before: main @ a3565ac | After: this PR @ de65a6b | |---|---|---| | three-areas · field-standing · frame rate | 23.9 | 23.8 (−0%) | | three-areas · kokiri-standing · frame rate | 23.5 | 24.5 (+4%) | | three-areas · market-standing · frame rate | 17.7 | 17.4 (−2%) | | three-areas · field-standing · stutters | 0.0 | 0.0 | | three-areas · kokiri-standing · stutters | 3.2 | 0.2 (−92%) | | three-areas · market-standing · stutters | 2.8 | 2.5 (−9%) | | three-areas · hitches over 50 ms in the whole run | 1,337 | 1,439 (+8%) | | three-areas · field-standing · frame time (p99) (ms) | 45.7 | 45.2 (−1%) | | three-areas · kokiri-standing · frame time (p99) (ms) | 72.0 | 49.6 (−31%) | | three-areas · market-standing · frame time (p99) (ms) | 97.0 | 93.4 (−4%) | ### Flags and settings None: the plan varied no setting, switch or build option. ``` </details> #### Measurements · amd+igpu **Verdict:** fail on amd+igpu (2/1 report): `three-areas screenshot.max_diff` failed. Hardware: 1: linux+amd+radv+igpu `AMD Radeon 780M Graphics (RADV PHOENIX), 26.1.99` `AMD Ryzen 7 7840HS w/ Radeon 780M Graphics, 24 GB RAM`; 2: linux+amd+radv+igpu `AMD Radeon 760M Graphics (RADV PHOENIX), 26.0.8` `AMD Ryzen 5 230 w/ Radeon 760M Graphics, 16 GB RAM` Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again) | | Before: main @ a3565ac | After: this PR @ de65a6b | |---|---|---| | three-areas · field-standing · frame rate | 33.4 | 33.3 (−0%) | | three-areas · kokiri-standing · frame rate | 35.9 | 37.0 (+3%) | | three-areas · market-standing · frame rate | 27.4 | 27.7 (+1%) | | three-areas · field-standing · stutters | 0.0 | 0.0 | | three-areas · kokiri-standing · stutters | 0.0 | 0.0 | | three-areas · market-standing · stutters | 4.0 | 0.2 (−94%) | | three-areas · hitches over 50 ms in the whole run | 1,014 | 1,018 (+0%) | | three-areas · field-standing · frame time (p99) (ms) | 52.7 | 53.2 (+1%) | | three-areas · kokiri-standing · frame time (p99) (ms) | 46.1 | 36.7 (−20%) | | three-areas · market-standing · frame time (p99) (ms) | 63.9 | 59.3 (−7%) | ### Flags and settings None: the plan varied no setting, switch or build option. <details><summary>Markdown for the PR description</summary> ```markdown ## Measurements **Verdict:** fail on amd+igpu (2/1 report): `three-areas screenshot.max_diff` failed. Hardware: 1: linux+amd+radv+igpu `AMD Radeon 780M Graphics (RADV PHOENIX), 26.1.99` `AMD Ryzen 7 7840HS w/ Radeon 780M Graphics, 24 GB RAM`; 2: linux+amd+radv+igpu `AMD Radeon 760M Graphics (RADV PHOENIX), 26.0.8` `AMD Ryzen 5 230 w/ Radeon 760M Graphics, 16 GB RAM` Scene: three-areas, from the repository's save kokiri-start (vsync=2, frame_rate_cap=6); each cell's settings and switches as the table names them Method: the Beanstalk test farm: main and this PR built and run on the same PC in one session, game starts alternating (ABBA); three-areas: 2 game starts a side, kokiri-standing 20 s after 5 s warm-up, field-standing 20 s after 5 s warm-up, market-standing 20 s after 5 s warm-up; each number the median of the runs, then of 2 reports How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again) | | Before: main @ a3565ac | After: this PR @ de65a6b | |---|---|---| | three-areas · field-standing · frame rate | 33.4 | 33.3 (−0%) | | three-areas · kokiri-standing · frame rate | 35.9 | 37.0 (+3%) | | three-areas · market-standing · frame rate | 27.4 | 27.7 (+1%) | | three-areas · field-standing · stutters | 0.0 | 0.0 | | three-areas · kokiri-standing · stutters | 0.0 | 0.0 | | three-areas · market-standing · stutters | 4.0 | 0.2 (−94%) | | three-areas · hitches over 50 ms in the whole run | 1,014 | 1,018 (+0%) | | three-areas · field-standing · frame time (p99) (ms) | 52.7 | 53.2 (+1%) | | three-areas · kokiri-standing · frame time (p99) (ms) | 46.1 | 36.7 (−20%) | | three-areas · market-standing · frame time (p99) (ms) | 63.9 | 59.3 (−7%) | ### Flags and settings None: the plan varied no setting, switch or build option. ``` </details> </details>
HumanBean changed title from WIP: perf(vk): a draw rebuilds only the descriptors whose sources changed (RX_VK_DESC_SPLIT) to perf(vk): a draw rebuilds only the descriptors whose sources changed (RX_VK_DESC_SPLIT) 2026-11-20 07:23:39 +00:00
HumanBean force-pushed perf/draw-cost from de65a6b467
Some checks failed
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been skipped
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been skipped
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been skipped
Third-party parts / As their notes say (pull_request) Successful in 14s
Third-party parts / The same as upstream (pull_request) Has been skipped
CI / Conventional Commit title (pull_request) Successful in 8s
Guide screenshots / Guide images and local UI source review (pull_request) Failing after 12s
CI / Workflow lint (actionlint) (pull_request) Successful in 13s
Website / Website on Cloudflare (pull_request_target) Has been skipped
PR description / Description follows the template (pull_request_target) Failing after 4s
beanstalk/windows+nvidia 1/1 fail
beanstalk/windows+intel 0/1 pass, 0 running, 2 queued
beanstalk/linux+amd 2/1 fail
beanstalk/linux+nvidia 2/1 fail
beanstalk/approval approved by HumanBean (reviewed)
beanstalk/windows+amd 2/1 fail
beanstalk/linux+intel 0/1 pass, 0 running, 2 queued
beanstalk/steamdeck 2/1 fail
beanstalk/amd+igpu 1/1 fail
CI / Website build and checks (pull_request) Has been cancelled
to 15b16be2a9
Some checks failed
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been skipped
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been skipped
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been skipped
Third-party parts / The same as upstream (pull_request) Has been skipped
Website / Website on Cloudflare (pull_request) Has been skipped
CI / Conventional Commit title (pull_request) Successful in 12s
PR description / Description follows the template (pull_request_target) Failing after 14s
Guide screenshots / Guide images and local UI source review (pull_request) Failing after 16s
CI / Workflow lint (actionlint) (pull_request) Successful in 17s
Third-party parts / As their notes say (pull_request) Successful in 18s
beanstalk/approval not approved yet
CI / Website build and checks (pull_request) Failing after 2m13s
2026-11-20 08:53:25 +00:00
Compare
Member

Beanstalk test farm

Commit 15b16be · tier untrusted · approved by nobody yet · dashboard

A member of BigBeans can review this commit and comment @Beanstalk to test it.

<!-- beanstalk:sticky --> ### Beanstalk test farm Commit `15b16be` · tier **untrusted** · approved by nobody yet · [dashboard](https://beanstalk.magicbean.dev/prs/1851) A member of `BigBeans` can review this commit and comment `@Beanstalk` to test it.
Author
Owner

Closing: the split is correct (RX_VK_DESC_SPLIT=2 compared kept textures and storage buffers with whole builds over 2.74 million draws on an RX 580 with 0 differences), but it gains nothing measurable. The Test Farm was even on 8 GPUs (within about 2%; the two large Hyrule Field drops compared runs that hadn't reached the same place), and on an RTX 3090 the frame's critical path in the Market is no longer the render thread, so cheaper descriptor builds don't move fps. Not worth a second descriptor path and a switch to maintain. Its counters (why a draw rebuilt its descriptors) were the useful part: they showed bindless textures and a texture memo wouldn't pay either. The branch stays on HumanBean/Longshot as perf/draw-cost.

Closing: the split is correct (`RX_VK_DESC_SPLIT=2` compared kept textures and storage buffers with whole builds over 2.74 million draws on an RX 580 with 0 differences), but it gains nothing measurable. The Test Farm was even on 8 GPUs (within about 2%; the two large Hyrule Field drops compared runs that hadn't reached the same place), and on an RTX 3090 the frame's critical path in the Market is no longer the render thread, so cheaper descriptor builds don't move fps. Not worth a second descriptor path and a switch to maintain. Its counters (why a draw rebuilt its descriptors) were the useful part: they showed bindless textures and a texture memo wouldn't pay either. The branch stays on HumanBean/Longshot as perf/draw-cost.
Author
Owner

Closing: correct but no measurable gain; reasons in the comment above.

Closing: correct but no measurable gain; reasons in the comment above.
HumanBean closed this pull request 2026-11-20 15:46:36 +00:00
Some checks failed
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been skipped
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been skipped
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been skipped
Third-party parts / The same as upstream (pull_request) Has been skipped
Website / Website on Cloudflare (pull_request) Has been skipped
CI / Conventional Commit title (pull_request) Successful in 12s
PR description / Description follows the template (pull_request_target) Failing after 14s
Guide screenshots / Guide images and local UI source review (pull_request) Failing after 16s
CI / Workflow lint (actionlint) (pull_request) Successful in 17s
Third-party parts / As their notes say (pull_request) Successful in 18s
beanstalk/approval not approved yet
CI / Website build and checks (pull_request) Failing after 2m13s

Pull request closed

Sign in to join this conversation.
No reviewers
No milestone
No project
No assignees
2 participants
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
MagicBeanDev/Longshot!1851
No description provided.