perf(host): default short sleep timer on Windows integrated GPUs to prevent APU TDP starvation #1114

Open
Jer9 wants to merge 2 commits from Jer9/Longshot:perf/apu-sleep-timer into main
First-time contributor

Type: perf · Risk: low · Tested on: Windows 11, AMD Ryzen AI 7 350 / Radeon 860M (2x2 interleaved A/B in ZoraDomain_Child + game-free C/Python checks) · Needs: Windows discrete GPU (> 6 threads); Linux, any GPU

Why

On multi-threaded Windows APUs (such as an 8-core / 16-thread AMD Ryzen AI 7 350 with Radeon 860M graphics), process_threads() > 6 caused host/hos_thread.c to leave how = 0 and set poll_cap = threads / 4 = 4 (from https://git.notrexed.dev/Hookshot/Hookshot/pulls/2709) when HOS_SHORT_SLEEP_TIMER is unset. OoT's idle guest job workers call 20 us sleeps thousands of times a second and spin in SwitchToThread() / YieldProcessor() loops across up to 4 CPU cores.

Because a mobile APU shares one package TDP between CPU cores and the integrated GPU, those 4 spinning pollers burn package power and throttle iGPU clocks. On Linux, https://git.notrexed.dev/Hookshot/Hookshot/pulls/2637 already defaulted how = 1 across all processor counts because OoT's idle job workers are not on the frame's critical path. Scoping the Windows how = 1 default strictly to integrated GPUs (VK_PHYSICAL_DEVICE_TYPE_INTEGRATED_GPU) lets guest short sleeps block on CreateWaitableTimerExW high-resolution timers and returns socket power headroom to the iGPU while keeping Windows discrete-GPU desktops completely unchanged on https://git.notrexed.dev/Hookshot/Hookshot/pulls/2709's processor threshold and HOS_SHORT_SLEEP_POLLERS cap. See #1040.

What changed

Adds an integrated-GPU query option (HOS_OPT_INTEGRATED_GPU = 62) on the existing renderer option callback so sleep_ns() in host/hos_thread.c defaults how = 1 on Windows only when the chosen Vulkan physical device is VK_PHYSICAL_DEVICE_TYPE_INTEGRATED_GPU, leaving discrete GPUs untouched.

  • vk / api: gpu/vk/vk_api.h defines HOS_OPT_INTEGRATED_GPU = 62 (answered by the existing option callback, so MkwVkHost, MkwVkRenderer, and MKWVK_API_VERSION are unchanged).
  • vk / replay: gpu/vk/vk_replay.cpp answers id == HOS_OPT_INTEGRATED_GPU in fill_replay_table()'s t.option lambda with dev.props.deviceType == VK_PHYSICAL_DEVICE_TYPE_INTEGRATED_GPU (populated during rx_init before nnMain or guest short sleeps start).
  • host: host/hos_thread.c declares int rx_option(int id, int otherwise);, updates #else (Windows) in sleep_ns() to set how = 1 when !getenv("HOS_SHORT_SLEEP_TIMER") && rx_option(62, 0), and updates the comments above hos_short_sleep_timer() and inside sleep_ns() to document the unset defaults across platforms, explicit =0/=1 (cap = 0), and sead::MainThread's 3.3–3.6 ms/frame readback yields. Discrete GPUs (rx_option(62, 0) == 0) continue using hos_short_sleep_timer() and HOS_SHORT_SLEEP_POLLERS.
  • docs: docs/development.md adds HOS_SHORT_SLEEP_TIMER=0|1, HOS_SHORT_SLEEP_POLLERS=<n> to ### Running a test; docs/tested/2026-11-16-short-sleep-igpu-jer9-old.md and docs/tested/2026-11-16-short-sleep-igpu-jer9-new.md record the 2x2 interleaved A/B runs on Windows 11 / Radeon 860M.
  • tests: tests/test_short_sleep_timer.py drops the duplicate pure-Python file (tests/test_short_sleep_default.py is removed), folds the HOS_OPT_INTEGRATED_GPU = 62 wiring checks against gpu/vk/vk_api.h and gpu/vk/vk_replay.cpp (t.option) directly into test_short_sleep_timer.py, stubs rx_option(62, 0) in its C fixture, adds igpu = 1 cases while preserving all main discrete-GPU cases, and adds igpu-ignored / dgpu-forced mutant checks.

Measurements

Verdict: Helps significantly on Windows multi-threaded APUs with HOS_SHORT_SLEEP_TIMER unset: across 2 interleaved compiled-binary A/B pairs in ZoraDomain_Child, Locked-View wall FPS improved +15.3% (24.32 -> 28.05 fps), settled log FPS improved +13.8% (25.0 -> 28.45 fps), GPU frame time dropped -22.3% (28.3 -> 22.0 ms), and > 40 ms present intervals fell from 32–76 to 3–4 per 5 s window with 0 texture budget evictions and 0 device losses. Auto-detection (how = 1 after Vulkan init) confirmed in longshot_log.txt.

Hardware: Windows 11 (build 26200), AMD Radeon(TM) 860M Graphics (integrated, 1002:1114, 8 CUs, RDNA 3.5, 12855 MB video memory), AMD proprietary driver 26.10.1 (LLPC), AMD Ryzen AI 7 350 (8c/16t, znver5), 32 GB RAM (26277 MB usable, 6 GB Variable Graphics Memory)
Scene: ZoraDomain_Child, 1920x1080 offscreen (scene 1728x1080, picture 2304x1440), Quality preset Minimum, Handheld profile (handheld_profile=1), game's own upscaler (upscaling=0), threaded_rendering=0, RX_VK_COMMIT_RESERVE_MB=512, RX_VK_ASTC_CPU=1, RX_VK_VRAM_MB=8192, HOS_SHORT_SLEEP_TIMER unset; stationary Locked View at pos=[4.61, 103.88, -36.14], cam=7.6 (1,840-1,855 draws/frame)
Method: Built both arms locally with nx2/build.py --standalone --kept-lift (llvm-mingw clang 23, RX_RELEASE=1; 0_main & 2_main built from main source 013b1d3a; 1_pr & 3_pr built from the Option B patch a9f9d4bd, committed cleanly as 3dde802). Ran 2 interleaved pairs (0_main, 1_pr, 2_main, 3_pr) via scripts/longshot_ctl.py --profile test --repo-save kokiri-start: teleport 13 --arrive 60, wait in_play, 180-present warm-up, then 360-present measured Locked View; per-run medians taken from the final three 0.0 uploads 120-present windows alongside wall clock 360 / elapsed_s.
How to reproduce: python -I scripts/longshot_ctl.py --run <dir> launch <GAME_DIR> --profile test --repo-save kokiri-start --repo . --env RX_VK_COMMIT_RESERVE_MB=512 --env RX_VK_ASTC_CPU=1 --env RX_VK_VRAM_MB=8192, teleport 13 --arrive 60, wait in_play, wait present --frames 180, wait present --frames 360

Before: main @ bbe421f After: this PR @ 3dde802 Δ
Locked-View wall FPS (360 presents; pair 1, pair 2 -> midpoint) 23.55, 25.09 fps (24.32 fps) 27.76, 28.34 fps (28.05 fps) +15.3% (+3.73 fps)
Settled log FPS (per-run median of last 3 windows -> midpoint) 24.8, 25.2 fps (25.0 fps) 28.0, 28.9 fps (28.45 fps) +13.8% (+3.45 fps)
Settled GPU frame time (per-run median -> midpoint) 28.7, 27.9 ms (28.3 ms) 22.5, 21.5 ms (22.0 ms) -22.3% (-6.3 ms)
Present intervals over 40 ms (last two 5 s windows per run) 56, 76 / 51, 32 4, 4 / 3, 3 ~93% fewer
Short-sleep dispatch (HOS_SHORT_SLEEP_TIMER unset) how = 0, cap = 4 (max 4 pollers) how = 1, cap = 0 (guest timer, host yield) Auto on iGPU
Texture budget evictions / device loss 0 / none 0 / none Unchanged

(Earlier single-run v3.0.0-rc.5 switch exploration on top of RX_VK_PASS_BARRIER=game with HOS_SHORT_SLEEP_TIMER=1 (how = 2) showed consistent behavior in the same scene: 24.10 -> 30.42 fps wall FPS, 28.5 -> 19.3 ms GPU time.)

0_main (main @ bbe421f, source 013b1d3a, HOS_SHORT_SLEEP_TIMER unset; rx: renderer line 55, sleep: line 80):

[hos MainThread] rx: renderer: Vulkan (longshot_vk.dll)
[hos host] sleep: at most 4 of the game's threads yield through a short sleep at a time, the others wait on the timer (16 logical processors; HOS_SHORT_SLEEP_POLLERS=0: all yield)
[hos Presentation Thread] vk: 3720 presents, 24.8 fps; a frame: 1848 draws (0 skipped), 129 dispatches, 160 passes, 162 barriers, 6 submits, 0.0 uploads (0.0 MB), 82 ub updates (57 overlays), GPU-bound wait 0.1 ms (recording slots only); 9314 pipelines (3847 ms), 0 programs (11 failed), 1721 textures (6037 MB)
[hos Presentation Thread] vk: GPU time 28.7 ms a frame (the longest submission 15.1 ms); 0.0 textures made a frame
[hos Presentation Thread] vi: present intervals: <15.5 ms 0, 15.5-17.8 0, 17.8-22 0, 22-30 0, 30-40 44, over 40 76

1_pr (this PR, source a9f9d4bd, HOS_SHORT_SLEEP_TIMER unset; rx: renderer line 55, sleep: line 80):

[hos MainThread] rx: renderer: Vulkan (longshot_vk.dll)
[hos host] sleep: the game's sleeps under a millisecond wait on the timer where the high-resolution one can be made (16 logical processors, 0: not counted; HOS_SHORT_SLEEP_TIMER=0: all yield)
[hos Presentation Thread] vk: 3960 presents, 28.0 fps; a frame: 1840 draws (0 skipped), 127 dispatches, 155 passes, 162 barriers, 6 submits, 0.0 uploads (0.0 MB), 82 ub updates (57 overlays), GPU-bound wait 0.1 ms (recording slots only); 9314 pipelines (2834 ms), 0 programs (11 failed), 1684 textures (5977 MB)
[hos Presentation Thread] vk: GPU time 22.6 ms a frame (the longest submission 13.2 ms); 0.0 textures made a frame
[hos Presentation Thread] vi: present intervals: <15.5 ms 0, 15.5-17.8 0, 17.8-22 0, 22-30 0, 30-40 137, over 40 4

2_main (main @ bbe421f, source 013b1d3a, HOS_SHORT_SLEEP_TIMER unset):

[hos host] sleep: at most 4 of the game's threads yield through a short sleep at a time, the others wait on the timer (16 logical processors; HOS_SHORT_SLEEP_POLLERS=0: all yield)
[hos Presentation Thread] vk: 3720 presents, 25.2 fps; a frame: 1855 draws (0 skipped), 128 dispatches, 159 passes, 162 barriers, 6 submits, 0.0 uploads (0.0 MB), 82 ub updates (57 overlays), GPU-bound wait 0.1 ms (recording slots only); 9314 pipelines (3025 ms), 0 programs (11 failed), 1747 textures (6078 MB)
[hos Presentation Thread] vk: GPU time 27.9 ms a frame (the longest submission 15.1 ms); 0.0 textures made a frame
[hos Presentation Thread] vi: present intervals: <15.5 ms 0, 15.5-17.8 0, 17.8-22 0, 22-30 0, 30-40 98, over 40 32

3_pr (this PR, source a9f9d4bd, HOS_SHORT_SLEEP_TIMER unset):

[hos host] sleep: the game's sleeps under a millisecond wait on the timer where the high-resolution one can be made (16 logical processors, 0: not counted; HOS_SHORT_SLEEP_TIMER=0: all yield)
[hos Presentation Thread] vk: 2520 presents, 28.9 fps; a frame: 1841 draws (0 skipped), 126 dispatches, 155 passes, 162 barriers, 6 submits, 0.0 uploads (0.0 MB), 82 ub updates (57 overlays), GPU-bound wait 0.1 ms (recording slots only); 9314 pipelines (3215 ms), 0 programs (11 failed), 1749 textures (6050 MB)
[hos Presentation Thread] vk: GPU time 21.5 ms a frame (the longest submission 11.4 ms); 0.0 textures made a frame
[hos Presentation Thread] vi: present intervals: <15.5 ms 0, 15.5-17.8 0, 17.8-22 0, 22-30 1, 30-40 141, over 40 3

Flags and settings

No new switch added; narrows the existing unset HOS_SHORT_SLEEP_TIMER default on Windows to integrated GPUs while keeping discrete GPUs on main's behavior, and adds HOS_SHORT_SLEEP_TIMER / HOS_SHORT_SLEEP_POLLERS to docs/development.md.

Flag What it is Before this PR After this PR Effect Helps / doesn't help / untested Recommendation
HOS_SHORT_SLEEP_TIMER How sleeps under 1 ms wait: 0 (yield all), 1 (guest on timer, host yields), 2 (all on timer) Unset: 0 on Windows with > 6 threads (cap = threads / 4), 1 on Linux Unset: 1 on Windows integrated GPUs (rx_option(62, 0)), 0 on Windows discrete GPUs (> 6 threads), 1 on Linux Guest short sleeps wait on high-resolution timer on Windows iGPUs (how = 1); host sleeps keep precision yield Helps on Windows APUs (+15.3% FPS, -22.3% GPU ms on Radeon 860M); Windows dGPU kept unchanged (untested on hardware here) on (how = 1) by default for Windows iGPUs

Testing

Tested:

  • Windows 11 (26200), AMD Ryzen AI 7 350 / Radeon 860M:
    • Game-free verification: 842dedb fails as expected on main (bbe421f); on 3dde802 python -I -B tests/test_short_sleep_timer.py --cc gcc (PASS), python -I -B tests/test_short_sleep_timer.py --cc clang (PASS), python -I -B tests/test_option_ids.py (PASS), python -I -B tests/test_doc_anchors.py (PASS), python -I -B tests/test_utf8_paths.py (PASS), python -I -B tests/test_spin_by_role.py (PASS), python -I -B tests/test_tested_configurations.py Repository (PASS), and python -I -B tests/run_all.py --exact tests/byte_compile_tracked.py tests/test_check_pr_description.py (2 passed, 0 failed).
    • Full compiled binary build (nx2/build.py --standalone --kept-lift, llvm-mingw clang 23) + 2x2 interleaved A/B benchmark in ZoraDomain_Child (longshot_ctl.py --profile test, HOS_SHORT_SLEEP_TIMER unset).
    • Suite runs: python -I -B tests/run_all.py --changed on 3dde802 (131 passed, 8 failed, 4 skipped with LONGSHOT_MINGW & RX_CPM, all 8 failures pre-existing Windows / main fixture drift: tests/test_crash_dialog.py, tests/test_fsr_uniform_ownership.py, tests/test_longshot_ctl.py, tests/test_pads_after_menu.py, tests/test_parallel_init_order.py, tests/test_replay_write_order.py, tests/test_vk_occlusion_queries.py, tests/test_walk_key.py), plus full python -I -B tests/run_all.py on the identical C/C++ implementation prior to dropping test_short_sleep_default.py and updating comments/docs/development.md (592 passed, 21 failed, 51 skipped without zstandard/dotnet/glslang/spirv-val, all 21 failures pre-existing Windows / main fixture drift unrelated to this PR: tests/test_auto_handheld.py, tests/test_bug_report_defaults.py, tests/test_crash_dialog.py, tests/test_crt_options.py, tests/test_fsr_uniform_ownership.py, tests/test_guest_imported.py, tests/test_import_reach.py, tests/test_linux_preflight.py, tests/test_longshot_ctl.py, tests/test_make_game_cache.py, tests/test_missing_nup.py, tests/test_pads_after_menu.py, tests/test_parallel_init_order.py, tests/test_pending_guest_import.py, tests/test_release.py, tests/test_replay_write_order.py, tests/test_run_all_changed.py, tests/test_srccut.py, tests/test_tested_configurations.py (CRLF pipe check only; Repository passes), tests/test_vk_occlusion_queries.py, tests/test_walk_key.py).

Not tested:

  • Windows discrete-GPU desktop builds (> 6 logical processors) on physical hardware (no dGPU in this laptop, though code path is unchanged when dev.props.deviceType != VK_PHYSICAL_DEVICE_TYPE_INTEGRATED_GPU).
  • Native Linux compile / full Linux tests/run_all.py (spirv-val, glslang, POSIX tests).

Must be tested:

  • Windows, discrete GPU (> 6 logical processors): run in play with HOS_SHORT_SLEEP_TIMER unset and inspect longshot_log.txt, expecting sleep: at most N of the game's threads yield through a short sleep at a time (unchanged from main, e.g. 4 on 16 threads) and identical FPS (Test Farm: @Beanstalk windows, or by hand on any dGPU desktop)
  • Linux, any GPU: run python -I -B tests/test_short_sleep_timer.py and python -I -B tests/run_all.py with gcc and clang, expecting all short-sleep and host checks to pass cleanly (Test Farm: @Beanstalk linux)

Risks

What could break: On older Windows builds lacking CREATE_WAITABLE_TIMER_HIGH_RESOLUTION (prior to Windows 10 1803), sleep_ns() falls back to SwitchToThread() yielding as before. On systems where an integrated GPU reports VK_PHYSICAL_DEVICE_TYPE_INTEGRATED_GPU, unset HOS_SHORT_SLEEP_TIMER switches guest short sleeps to the high-resolution timer (how = 1). Discrete GPUs (VK_PHYSICAL_DEVICE_TYPE_DISCRETE_GPU) do not enter this path.

Turning it off: Set HOS_SHORT_SLEEP_TIMER=0 in longshot_env.txt or revert commit 3dde802.

Compatibility: Nothing changes for saves, longshot_options.ini, shader/pipeline caches, or MKWVK_API_VERSION (adding HOS_OPT_INTEGRATED_GPU = 62 to the existing option callback does not change MkwVkHost or MkwVkRenderer).

Issues

See #1040

Checklist

Everyone

  • One topic.
  • The title is a Conventional Commit (type(scope): description) of the status line's type.
  • Why gives the reason plainly and links the issues that fit (the tracker searched).
  • Measurements has a verdict, or one line saying why there is nothing to measure.
  • Testing names what was tested, what wasn't, and what must still be tested, where and how.
  • Rebased on current main.
  • python -I -B tests/run_all.py passes (any failure that also fails on main is named).
  • No game files, dumps, build output or generated files.
  • third_party/ untouched (a new folder with a PROVENANCE.md instead).
  • The license is OK: GPL-3.0-or-later, nothing incompatible and nothing decompiled.

When it applies

  • AI assistance recorded with Assisted-by: / Co-Authored-By: trailers.
  • Numbers from the same PC, scene and settings, at least 2 interleaved runs each, the median reported.
  • Switches documented (docs/development.md; the README for a menu row) and shipped at their recommended value.
  • A file in docs/tested/ for each new measured run (format in docs/tested-configurations.md).
  • A game-free test added for the fix, failing on main and passing here.
  • A check under tests/ that reads changed source updated in this PR, not weakened.
  • Docs updated.
**Type:** perf · **Risk:** low · **Tested on:** Windows 11, AMD Ryzen AI 7 350 / Radeon 860M (2x2 interleaved A/B in ZoraDomain_Child + game-free C/Python checks) · **Needs:** Windows discrete GPU (> 6 threads); Linux, any GPU ## Why On multi-threaded Windows APUs (such as an 8-core / 16-thread AMD Ryzen AI 7 350 with Radeon 860M graphics), `process_threads() > 6` caused `host/hos_thread.c` to leave `how = 0` and set `poll_cap = threads / 4 = 4` (from https://git.notrexed.dev/Hookshot/Hookshot/pulls/2709) when `HOS_SHORT_SLEEP_TIMER` is unset. OoT's idle guest job workers call 20 us sleeps thousands of times a second and spin in `SwitchToThread()` / `YieldProcessor()` loops across up to 4 CPU cores. Because a mobile APU shares one package TDP between CPU cores and the integrated GPU, those 4 spinning pollers burn package power and throttle iGPU clocks. On Linux, https://git.notrexed.dev/Hookshot/Hookshot/pulls/2637 already defaulted `how = 1` across all processor counts because OoT's idle job workers are not on the frame's critical path. Scoping the Windows `how = 1` default strictly to integrated GPUs (`VK_PHYSICAL_DEVICE_TYPE_INTEGRATED_GPU`) lets guest short sleeps block on `CreateWaitableTimerExW` high-resolution timers and returns socket power headroom to the iGPU while keeping Windows discrete-GPU desktops completely unchanged on https://git.notrexed.dev/Hookshot/Hookshot/pulls/2709's processor threshold and `HOS_SHORT_SLEEP_POLLERS` cap. See #1040. ## What changed Adds an integrated-GPU query option (`HOS_OPT_INTEGRATED_GPU = 62`) on the existing renderer option callback so `sleep_ns()` in `host/hos_thread.c` defaults `how = 1` on Windows only when the chosen Vulkan physical device is `VK_PHYSICAL_DEVICE_TYPE_INTEGRATED_GPU`, leaving discrete GPUs untouched. - vk / api: `gpu/vk/vk_api.h` defines `HOS_OPT_INTEGRATED_GPU = 62` (answered by the existing `option` callback, so `MkwVkHost`, `MkwVkRenderer`, and `MKWVK_API_VERSION` are unchanged). - vk / replay: `gpu/vk/vk_replay.cpp` answers `id == HOS_OPT_INTEGRATED_GPU` in `fill_replay_table()`'s `t.option` lambda with `dev.props.deviceType == VK_PHYSICAL_DEVICE_TYPE_INTEGRATED_GPU` (populated during `rx_init` before `nnMain` or guest short sleeps start). - host: `host/hos_thread.c` declares `int rx_option(int id, int otherwise);`, updates `#else` (Windows) in `sleep_ns()` to set `how = 1` when `!getenv("HOS_SHORT_SLEEP_TIMER") && rx_option(62, 0)`, and updates the comments above `hos_short_sleep_timer()` and inside `sleep_ns()` to document the unset defaults across platforms, explicit `=0`/`=1` (`cap = 0`), and `sead::MainThread`'s 3.3–3.6 ms/frame readback yields. Discrete GPUs (`rx_option(62, 0) == 0`) continue using `hos_short_sleep_timer()` and `HOS_SHORT_SLEEP_POLLERS`. - docs: `docs/development.md` adds `HOS_SHORT_SLEEP_TIMER=0|1`, `HOS_SHORT_SLEEP_POLLERS=<n>` to `### Running a test`; `docs/tested/2026-11-16-short-sleep-igpu-jer9-old.md` and `docs/tested/2026-11-16-short-sleep-igpu-jer9-new.md` record the 2x2 interleaved A/B runs on Windows 11 / Radeon 860M. - tests: `tests/test_short_sleep_timer.py` drops the duplicate pure-Python file (`tests/test_short_sleep_default.py` is removed), folds the `HOS_OPT_INTEGRATED_GPU = 62` wiring checks against `gpu/vk/vk_api.h` and `gpu/vk/vk_replay.cpp` (`t.option`) directly into `test_short_sleep_timer.py`, stubs `rx_option(62, 0)` in its C fixture, adds `igpu = 1` cases while preserving all `main` discrete-GPU cases, and adds `igpu-ignored` / `dgpu-forced` mutant checks. ## Measurements **Verdict:** Helps significantly on Windows multi-threaded APUs with `HOS_SHORT_SLEEP_TIMER` unset: across 2 interleaved compiled-binary A/B pairs in `ZoraDomain_Child`, Locked-View wall FPS improved **+15.3%** (`24.32 -> 28.05 fps`), settled log FPS improved **+13.8%** (`25.0 -> 28.45 fps`), GPU frame time dropped **-22.3%** (`28.3 -> 22.0 ms`), and > 40 ms present intervals fell from `32–76` to `3–4` per 5 s window with `0` texture budget evictions and `0` device losses. Auto-detection (`how = 1` after Vulkan init) confirmed in `longshot_log.txt`. Hardware: Windows 11 (build 26200), AMD Radeon(TM) 860M Graphics (integrated, 1002:1114, 8 CUs, RDNA 3.5, 12855 MB video memory), AMD proprietary driver 26.10.1 (LLPC), AMD Ryzen AI 7 350 (8c/16t, znver5), 32 GB RAM (26277 MB usable, 6 GB Variable Graphics Memory) Scene: `ZoraDomain_Child`, `1920x1080` offscreen (scene `1728x1080`, picture `2304x1440`), Quality preset Minimum, Handheld profile (`handheld_profile=1`), game's own upscaler (`upscaling=0`), `threaded_rendering=0`, `RX_VK_COMMIT_RESERVE_MB=512`, `RX_VK_ASTC_CPU=1`, `RX_VK_VRAM_MB=8192`, `HOS_SHORT_SLEEP_TIMER` unset; stationary Locked View at `pos=[4.61, 103.88, -36.14]`, `cam=7.6` (`1,840-1,855` draws/frame) Method: Built both arms locally with `nx2/build.py --standalone --kept-lift` (llvm-mingw clang 23, `RX_RELEASE=1`; `0_main` & `2_main` built from `main` source `013b1d3a`; `1_pr` & `3_pr` built from the Option B patch `a9f9d4bd`, committed cleanly as `3dde802`). Ran 2 interleaved pairs (`0_main`, `1_pr`, `2_main`, `3_pr`) via `scripts/longshot_ctl.py --profile test --repo-save kokiri-start`: `teleport 13 --arrive 60`, `wait in_play`, 180-present warm-up, then 360-present measured Locked View; per-run medians taken from the final three `0.0 uploads` 120-present windows alongside wall clock `360 / elapsed_s`. How to reproduce: `python -I scripts/longshot_ctl.py --run <dir> launch <GAME_DIR> --profile test --repo-save kokiri-start --repo . --env RX_VK_COMMIT_RESERVE_MB=512 --env RX_VK_ASTC_CPU=1 --env RX_VK_VRAM_MB=8192`, `teleport 13 --arrive 60`, `wait in_play`, `wait present --frames 180`, `wait present --frames 360` | | Before: main @ bbe421f | After: this PR @ 3dde802 | Δ | |---|---|---|---| | Locked-View wall FPS (360 presents; pair 1, pair 2 -> midpoint) | `23.55`, `25.09` fps (`24.32 fps`) | `27.76`, `28.34` fps (`28.05 fps`) | **+15.3%** (+3.73 fps) | | Settled log FPS (per-run median of last 3 windows -> midpoint) | `24.8`, `25.2` fps (`25.0 fps`) | `28.0`, `28.9` fps (`28.45 fps`) | **+13.8%** (+3.45 fps) | | Settled GPU frame time (per-run median -> midpoint) | `28.7`, `27.9` ms (`28.3 ms`) | `22.5`, `21.5` ms (`22.0 ms`) | **-22.3%** (-6.3 ms) | | Present intervals over 40 ms (last two 5 s windows per run) | `56, 76` / `51, 32` | `4, 4` / `3, 3` | **~93% fewer** | | Short-sleep dispatch (`HOS_SHORT_SLEEP_TIMER` unset) | `how = 0, cap = 4` (max 4 pollers) | `how = 1, cap = 0` (guest timer, host yield) | Auto on iGPU | | Texture budget evictions / device loss | `0` / none | `0` / none | Unchanged | *(Earlier single-run `v3.0.0-rc.5` switch exploration on top of `RX_VK_PASS_BARRIER=game` with `HOS_SHORT_SLEEP_TIMER=1` (`how = 2`) showed consistent behavior in the same scene: `24.10 -> 30.42 fps` wall FPS, `28.5 -> 19.3 ms` GPU time.)* `0_main` (`main` @ `bbe421f`, source `013b1d3a`, `HOS_SHORT_SLEEP_TIMER` unset; `rx: renderer` line 55, `sleep:` line 80): ``` [hos MainThread] rx: renderer: Vulkan (longshot_vk.dll) [hos host] sleep: at most 4 of the game's threads yield through a short sleep at a time, the others wait on the timer (16 logical processors; HOS_SHORT_SLEEP_POLLERS=0: all yield) [hos Presentation Thread] vk: 3720 presents, 24.8 fps; a frame: 1848 draws (0 skipped), 129 dispatches, 160 passes, 162 barriers, 6 submits, 0.0 uploads (0.0 MB), 82 ub updates (57 overlays), GPU-bound wait 0.1 ms (recording slots only); 9314 pipelines (3847 ms), 0 programs (11 failed), 1721 textures (6037 MB) [hos Presentation Thread] vk: GPU time 28.7 ms a frame (the longest submission 15.1 ms); 0.0 textures made a frame [hos Presentation Thread] vi: present intervals: <15.5 ms 0, 15.5-17.8 0, 17.8-22 0, 22-30 0, 30-40 44, over 40 76 ``` `1_pr` (`this PR`, source `a9f9d4bd`, `HOS_SHORT_SLEEP_TIMER` unset; `rx: renderer` line 55, `sleep:` line 80): ``` [hos MainThread] rx: renderer: Vulkan (longshot_vk.dll) [hos host] sleep: the game's sleeps under a millisecond wait on the timer where the high-resolution one can be made (16 logical processors, 0: not counted; HOS_SHORT_SLEEP_TIMER=0: all yield) [hos Presentation Thread] vk: 3960 presents, 28.0 fps; a frame: 1840 draws (0 skipped), 127 dispatches, 155 passes, 162 barriers, 6 submits, 0.0 uploads (0.0 MB), 82 ub updates (57 overlays), GPU-bound wait 0.1 ms (recording slots only); 9314 pipelines (2834 ms), 0 programs (11 failed), 1684 textures (5977 MB) [hos Presentation Thread] vk: GPU time 22.6 ms a frame (the longest submission 13.2 ms); 0.0 textures made a frame [hos Presentation Thread] vi: present intervals: <15.5 ms 0, 15.5-17.8 0, 17.8-22 0, 22-30 0, 30-40 137, over 40 4 ``` `2_main` (`main` @ `bbe421f`, source `013b1d3a`, `HOS_SHORT_SLEEP_TIMER` unset): ``` [hos host] sleep: at most 4 of the game's threads yield through a short sleep at a time, the others wait on the timer (16 logical processors; HOS_SHORT_SLEEP_POLLERS=0: all yield) [hos Presentation Thread] vk: 3720 presents, 25.2 fps; a frame: 1855 draws (0 skipped), 128 dispatches, 159 passes, 162 barriers, 6 submits, 0.0 uploads (0.0 MB), 82 ub updates (57 overlays), GPU-bound wait 0.1 ms (recording slots only); 9314 pipelines (3025 ms), 0 programs (11 failed), 1747 textures (6078 MB) [hos Presentation Thread] vk: GPU time 27.9 ms a frame (the longest submission 15.1 ms); 0.0 textures made a frame [hos Presentation Thread] vi: present intervals: <15.5 ms 0, 15.5-17.8 0, 17.8-22 0, 22-30 0, 30-40 98, over 40 32 ``` `3_pr` (`this PR`, source `a9f9d4bd`, `HOS_SHORT_SLEEP_TIMER` unset): ``` [hos host] sleep: the game's sleeps under a millisecond wait on the timer where the high-resolution one can be made (16 logical processors, 0: not counted; HOS_SHORT_SLEEP_TIMER=0: all yield) [hos Presentation Thread] vk: 2520 presents, 28.9 fps; a frame: 1841 draws (0 skipped), 126 dispatches, 155 passes, 162 barriers, 6 submits, 0.0 uploads (0.0 MB), 82 ub updates (57 overlays), GPU-bound wait 0.1 ms (recording slots only); 9314 pipelines (3215 ms), 0 programs (11 failed), 1749 textures (6050 MB) [hos Presentation Thread] vk: GPU time 21.5 ms a frame (the longest submission 11.4 ms); 0.0 textures made a frame [hos Presentation Thread] vi: present intervals: <15.5 ms 0, 15.5-17.8 0, 17.8-22 0, 22-30 1, 30-40 141, over 40 3 ``` ### Flags and settings No new switch added; narrows the existing unset `HOS_SHORT_SLEEP_TIMER` default on Windows to integrated GPUs while keeping discrete GPUs on `main`'s behavior, and adds `HOS_SHORT_SLEEP_TIMER` / `HOS_SHORT_SLEEP_POLLERS` to `docs/development.md`. | Flag | What it is | Before this PR | After this PR | Effect | Helps / doesn't help / untested | Recommendation | |---|---|---|---|---|---|---| | `HOS_SHORT_SLEEP_TIMER` | How sleeps under 1 ms wait: `0` (yield all), `1` (guest on timer, host yields), `2` (all on timer) | Unset: `0` on Windows with > 6 threads (`cap = threads / 4`), `1` on Linux | Unset: `1` on Windows integrated GPUs (`rx_option(62, 0)`), `0` on Windows discrete GPUs (> 6 threads), `1` on Linux | Guest short sleeps wait on high-resolution timer on Windows iGPUs (`how = 1`); host sleeps keep precision yield | Helps on Windows APUs (`+15.3%` FPS, `-22.3%` GPU ms on Radeon 860M); Windows dGPU kept unchanged (untested on hardware here) | on (`how = 1`) by default for Windows iGPUs | ## Testing **Tested:** - Windows 11 (26200), AMD Ryzen AI 7 350 / Radeon 860M: - Game-free verification: `842dedb` fails as expected on `main` (`bbe421f`); on `3dde802` `python -I -B tests/test_short_sleep_timer.py --cc gcc` (PASS), `python -I -B tests/test_short_sleep_timer.py --cc clang` (PASS), `python -I -B tests/test_option_ids.py` (PASS), `python -I -B tests/test_doc_anchors.py` (PASS), `python -I -B tests/test_utf8_paths.py` (PASS), `python -I -B tests/test_spin_by_role.py` (PASS), `python -I -B tests/test_tested_configurations.py Repository` (PASS), and `python -I -B tests/run_all.py --exact tests/byte_compile_tracked.py tests/test_check_pr_description.py` (2 passed, 0 failed). - Full compiled binary build (`nx2/build.py --standalone --kept-lift`, llvm-mingw clang 23) + 2x2 interleaved A/B benchmark in `ZoraDomain_Child` (`longshot_ctl.py --profile test`, `HOS_SHORT_SLEEP_TIMER` unset). - Suite runs: `python -I -B tests/run_all.py --changed` on `3dde802` (`131 passed, 8 failed, 4 skipped` with `LONGSHOT_MINGW` & `RX_CPM`, all 8 failures pre-existing Windows / `main` fixture drift: `tests/test_crash_dialog.py`, `tests/test_fsr_uniform_ownership.py`, `tests/test_longshot_ctl.py`, `tests/test_pads_after_menu.py`, `tests/test_parallel_init_order.py`, `tests/test_replay_write_order.py`, `tests/test_vk_occlusion_queries.py`, `tests/test_walk_key.py`), plus full `python -I -B tests/run_all.py` on the identical C/C++ implementation prior to dropping `test_short_sleep_default.py` and updating comments/`docs/development.md` (`592 passed, 21 failed, 51 skipped` without `zstandard`/`dotnet`/`glslang`/`spirv-val`, all 21 failures pre-existing Windows / `main` fixture drift unrelated to this PR: `tests/test_auto_handheld.py`, `tests/test_bug_report_defaults.py`, `tests/test_crash_dialog.py`, `tests/test_crt_options.py`, `tests/test_fsr_uniform_ownership.py`, `tests/test_guest_imported.py`, `tests/test_import_reach.py`, `tests/test_linux_preflight.py`, `tests/test_longshot_ctl.py`, `tests/test_make_game_cache.py`, `tests/test_missing_nup.py`, `tests/test_pads_after_menu.py`, `tests/test_parallel_init_order.py`, `tests/test_pending_guest_import.py`, `tests/test_release.py`, `tests/test_replay_write_order.py`, `tests/test_run_all_changed.py`, `tests/test_srccut.py`, `tests/test_tested_configurations.py` (CRLF pipe check only; `Repository` passes), `tests/test_vk_occlusion_queries.py`, `tests/test_walk_key.py`). **Not tested:** - Windows discrete-GPU desktop builds (> 6 logical processors) on physical hardware (no dGPU in this laptop, though code path is unchanged when `dev.props.deviceType != VK_PHYSICAL_DEVICE_TYPE_INTEGRATED_GPU`). - Native Linux compile / full Linux `tests/run_all.py` (`spirv-val`, `glslang`, POSIX tests). **Must be tested:** - Windows, discrete GPU (> 6 logical processors): run in play with `HOS_SHORT_SLEEP_TIMER` unset and inspect `longshot_log.txt`, expecting `sleep: at most N of the game's threads yield through a short sleep at a time` (unchanged from `main`, e.g. `4` on 16 threads) and identical FPS (Test Farm: `@Beanstalk windows`, or by hand on any dGPU desktop) - Linux, any GPU: run `python -I -B tests/test_short_sleep_timer.py` and `python -I -B tests/run_all.py` with `gcc` and `clang`, expecting all short-sleep and host checks to pass cleanly (Test Farm: `@Beanstalk linux`) ## Risks **What could break:** On older Windows builds lacking `CREATE_WAITABLE_TIMER_HIGH_RESOLUTION` (prior to Windows 10 1803), `sleep_ns()` falls back to `SwitchToThread()` yielding as before. On systems where an integrated GPU reports `VK_PHYSICAL_DEVICE_TYPE_INTEGRATED_GPU`, unset `HOS_SHORT_SLEEP_TIMER` switches guest short sleeps to the high-resolution timer (`how = 1`). Discrete GPUs (`VK_PHYSICAL_DEVICE_TYPE_DISCRETE_GPU`) do not enter this path. **Turning it off:** Set `HOS_SHORT_SLEEP_TIMER=0` in `longshot_env.txt` or revert commit `3dde802`. **Compatibility:** Nothing changes for saves, `longshot_options.ini`, shader/pipeline caches, or `MKWVK_API_VERSION` (adding `HOS_OPT_INTEGRATED_GPU = 62` to the existing option callback does not change `MkwVkHost` or `MkwVkRenderer`). ## Issues See #1040 ## Checklist ### Everyone - [x] One topic. - [x] The title is a Conventional Commit (`type(scope): description`) of the status line's type. - [x] Why gives the reason plainly and links the issues that fit (the tracker searched). - [x] Measurements has a verdict, or one line saying why there is nothing to measure. - [x] Testing names what was tested, what wasn't, and what must still be tested, where and how. - [x] Rebased on current `main`. - [x] `python -I -B tests/run_all.py` passes (any failure that also fails on `main` is named). - [x] No game files, dumps, build output or generated files. - [x] `third_party/` untouched (a new folder with a `PROVENANCE.md` instead). - [x] The license is OK: GPL-3.0-or-later, nothing incompatible and nothing decompiled. ### When it applies - [x] AI assistance recorded with `Assisted-by:` / `Co-Authored-By:` trailers. - [x] Numbers from the same PC, scene and settings, at least 2 interleaved runs each, the median reported. - [x] Switches documented (docs/development.md; the README for a menu row) and shipped at their recommended value. - [x] A file in `docs/tested/` for each new measured run (format in `docs/tested-configurations.md`). - [x] A game-free test added for the fix, failing on `main` and passing here. - [x] A check under `tests/` that reads changed source updated in this PR, not weakened. - [x] Docs updated.
Assisted-by: Antigravity
perf(host): default short sleep timer on Windows to prevent APU TDP starvation
Some checks failed
Website / Website on Cloudflare (pull_request_target) Has been skipped
PR description / Description follows the template (pull_request_target) Successful in 5s
CI / Conventional Commit title (pull_request) Has been cancelled
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been cancelled
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been cancelled
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been cancelled
CI / Workflow lint (actionlint) (pull_request) Has been cancelled
Guide screenshots / Guide images and local UI source review (pull_request) Has been cancelled
Third-party parts / As their notes say (pull_request) Has been cancelled
Third-party parts / The same as upstream (pull_request) Has been cancelled
2b34e076c2
Assisted-by: Antigravity
Author
First-time contributor

Tested in game: Windows 11, AMD Radeon 860M, Adrenalin 26.10.1: ZoraDomain_Child locked view on v3.0.0-rc.5 (+22% FPS, -31.3% GPU frame time); game-free tests pass on branch

Tested in game: Windows 11, AMD Radeon 860M, Adrenalin 26.10.1: ZoraDomain_Child locked view on v3.0.0-rc.5 (+22% FPS, -31.3% GPU frame time); game-free tests pass on branch
dyst left a comment

Small, easy-to-revert change, and the APU TDP story is believable. A few things before it becomes the default for every Windows desktop:

  1. Reach vs evidence. Windows with 6 or fewer threads already gets how = 1, so this only changes PCs with 7+ logical processors, which is nearly every desktop. The evidence is one APU, measuring HOS_SHORT_SLEEP_TIMER=1 (how = 2, not the how = 1 shipped). Those are single runs on rc.5, with RX_VK_PASS_BARRIER=game only on the "after" side.
  2. The serial threads lose their exemption. On main, MainThread and the Presentation Thread always yield (line 1205). Under how = 1 their short sleeps go to the Windows timer, about 0.5 ms late (line 1104), on the frame's critical path. The Linux evidence doesn't carry over, because the shim waits with clock_nanosleep (win32_posix.c:530). Your POLLERS=1 column was the fastest, so a lower cap, or the timer with the serial threads still yielding, may keep the win with less risk.
  3. Stale docs. The comments at lines 1103-1114 and 1195, and the log line, still describe the old defaults (=0 now gets the cap). HOS_SHORT_SLEEP_TIMER also has no docs/development.md row, and this is a good time to add one.
  4. test_short_sleep_default.py copies hos_short_sleep_pollers into Python (453-464) instead of reading it. test_short_sleep_timer.py already compiles the real code, so I'd drop it.

This is from reading the code only: nothing was built or run.

🤖 Generated with Claude Code

Small, easy-to-revert change, and the APU TDP story is believable. A few things before it becomes the default for every Windows desktop: 1. **Reach vs evidence.** Windows with 6 or fewer threads already gets `how = 1`, so this only changes PCs with 7+ logical processors, which is nearly every desktop. The evidence is one APU, measuring `HOS_SHORT_SLEEP_TIMER=1` (`how = 2`, not the `how = 1` shipped). Those are single runs on rc.5, with `RX_VK_PASS_BARRIER=game` only on the "after" side. 2. **The serial threads lose their exemption.** On main, MainThread and the Presentation Thread always yield (line 1205). Under `how = 1` their short sleeps go to the Windows timer, about 0.5 ms late (line 1104), on the frame's critical path. The Linux evidence doesn't carry over, because the shim waits with `clock_nanosleep` (`win32_posix.c:530`). Your `POLLERS=1` column was the fastest, so a lower cap, or the timer with the serial threads still yielding, may keep the win with less risk. 3. **Stale docs.** The comments at lines 1103-1114 and 1195, and the log line, still describe the old defaults (`=0` now gets the cap). `HOS_SHORT_SLEEP_TIMER` also has no docs/development.md row, and this is a good time to add one. 4. `test_short_sleep_default.py` copies `hos_short_sleep_pollers` into Python (453-464) instead of reading it. `test_short_sleep_timer.py` already compiles the real code, so I'd drop it. This is from reading the code only: nothing was built or run. 🤖 Generated with [Claude Code](https://claude.com/claude-code)
@ -1191,1 +1184,4 @@
the host's still yield (how = 1). On multi-threaded Windows APUs, spinning idle worker pollers starve
the integrated GPU of package TDP; on Linux (#2602), OoT's idle job workers yielded ~9 ms of CPU a frame
without being on the critical path. HOS_SHORT_SLEEP_TIMER=0: all yield, as before; =1: all on the timer */
if (!getenv("HOS_SHORT_SLEEP_TIMER")) how = 1;
Member

With how = 1, short_timer < (guest ? 1 : 2) is false for every guest sleep, so sead::MainThread and the Presentation Thread go to the timer as well. On main above 6 threads they always yield (!hos_thread_serial(hos_self()) at 1205). The removed Linux comment notes that MainThread itself makes these short sleeps. On Windows that is about 0.5 ms late per wake (line 1104). On Linux it is clock_nanosleep (host/posix/win32_posix.c:530), so the Linux result doesn't tell us much here. Could the serial threads keep yielding under how = 1?

With `how = 1`, `short_timer < (guest ? 1 : 2)` is false for every guest sleep, so sead::MainThread and the Presentation Thread go to the timer as well. On main above 6 threads they always yield (`!hos_thread_serial(hos_self())` at 1205). The removed Linux comment notes that MainThread itself makes these short sleeps. On Windows that is about 0.5 ms late per wake (line 1104). On Linux it is `clock_nanosleep` (host/posix/win32_posix.c:530), so the Linux result doesn't tell us much here. Could the serial threads keep yielding under `how = 1`?
@ -1198,7 +1193,7 @@ static void sleep_ns(int64_t ns, int guest) {
"(%d logical processors, 0: not counted; HOS_SHORT_SLEEP_TIMER=0: all yield)",
how == 2 ? "all" : "the game's", threads);
/* (HOS_SHORT_SLEEP_TIMER set says how all short sleeps wait: no limit then) */
Member

This comment is stale now: an explicit HOS_SHORT_SLEEP_TIMER=0 gets the poller cap. The same goes for the log line's "HOS_SHORT_SLEEP_TIMER=0: all yield" just above, and for hos_short_sleep_timer's comment (1103-1114), which still describes the unset default by processor count.

This comment is stale now: an explicit `HOS_SHORT_SLEEP_TIMER=0` gets the poller cap. The same goes for the log line's "HOS_SHORT_SLEEP_TIMER=0: all yield" just above, and for `hos_short_sleep_timer`'s comment (1103-1114), which still describes the unset default by processor count.
@ -0,0 +450,4 @@
"SHORT_SLEEP_TIMER_THREADS": short_sleep_timer_threads,
}))
# Fallback for hos_short_sleep_pollers
if "strtol" in body:
Member

This fallback is a Python copy of hos_short_sleep_pollers, not the source. If the C changes, this keeps passing. tests/test_short_sleep_timer.py already compiles the real sleep_ns / choice / pollers and covers these cases, so I'd drop this file rather than maintain a 560-line C evaluator.

This fallback is a Python copy of `hos_short_sleep_pollers`, not the source. If the C changes, this keeps passing. `tests/test_short_sleep_timer.py` already compiles the real `sleep_ns` / choice / pollers and covers these cases, so I'd drop this file rather than maintain a 560-line C evaluator.
Jer9 force-pushed perf/apu-sleep-timer from 2b34e076c2
Some checks failed
Website / Website on Cloudflare (pull_request_target) Has been skipped
PR description / Description follows the template (pull_request_target) Successful in 5s
CI / Conventional Commit title (pull_request) Has been cancelled
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been cancelled
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been cancelled
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been cancelled
CI / Workflow lint (actionlint) (pull_request) Has been cancelled
Guide screenshots / Guide images and local UI source review (pull_request) Has been cancelled
Third-party parts / As their notes say (pull_request) Has been cancelled
Third-party parts / The same as upstream (pull_request) Has been cancelled
to e2011b2761
Some checks failed
PR description / Description follows the template (pull_request_target) Successful in 4s
CI / Conventional Commit title (pull_request) Has been cancelled
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been cancelled
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been cancelled
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been cancelled
CI / Workflow lint (actionlint) (pull_request) Has been cancelled
Guide screenshots / Guide images and local UI source review (pull_request) Has been cancelled
Third-party parts / As their notes say (pull_request) Has been cancelled
Third-party parts / The same as upstream (pull_request) Has been cancelled
Website / Website on Cloudflare (pull_request) Has been cancelled
2026-11-16 12:50:03 +00:00
Compare
Author
First-time contributor

Switched to Option B (scoping the Windows how = 1 default to integrated GPUs via HOS_OPT_INTEGRATED_GPU = 62 so discrete-GPU desktops keep #2709's processor rule and HOS_SHORT_SLEEP_POLLERS), added 2x2 interleaved compiled-binary A/B measurements on Ryzen AI 7 350 / Radeon 860M (+15.3% wall FPS, -22.3% GPU frame time), and added docs/tested/ fragments.

Switched to Option B (scoping the Windows how = 1 default to integrated GPUs via HOS_OPT_INTEGRATED_GPU = 62 so discrete-GPU desktops keep #2709's processor rule and HOS_SHORT_SLEEP_POLLERS), added 2x2 interleaved compiled-binary A/B measurements on Ryzen AI 7 350 / Radeon 860M (+15.3% wall FPS, -22.3% GPU frame time), and added docs/tested/ fragments.
Jer9 changed title from WIP: perf(host): default short sleep timer on Windows to prevent APU TDP starvation to perf(host): default short sleep timer on Windows to prevent APU TDP starvation 2026-11-16 12:50:42 +00:00
Jer9 changed title from perf(host): default short sleep timer on Windows to prevent APU TDP starvation to perf(host): default short sleep timer on Windows integrated GPUs to prevent APU TDP starvation 2026-11-16 13:28:29 +00:00
Jer9 force-pushed perf/apu-sleep-timer from e2011b2761
Some checks failed
PR description / Description follows the template (pull_request_target) Successful in 4s
CI / Conventional Commit title (pull_request) Has been cancelled
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been cancelled
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been cancelled
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been cancelled
CI / Workflow lint (actionlint) (pull_request) Has been cancelled
Guide screenshots / Guide images and local UI source review (pull_request) Has been cancelled
Third-party parts / As their notes say (pull_request) Has been cancelled
Third-party parts / The same as upstream (pull_request) Has been cancelled
Website / Website on Cloudflare (pull_request) Has been cancelled
to 3dde802f47
Some checks reported errors
CI / Conventional Commit title (pull_request) Blocked by required conditions
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Blocked by required conditions
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Blocked by required conditions
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Blocked by required conditions
Guide screenshots / Guide images and local UI source review (pull_request) Blocked by required conditions
Third-party parts / As their notes say (pull_request) Blocked by required conditions
Third-party parts / The same as upstream (pull_request) Blocked by required conditions
CI / Workflow lint (actionlint) (pull_request) Blocked by required conditions
Website / Website on Cloudflare (pull_request) Blocked by required conditions
PR description / Description follows the template (pull_request_target) Successful in 4s
Website / Website on Cloudflare (pull_request_target) Has been skipped
beanstalk/approval approved by HumanBean (reviewed)
beanstalk/windows+igpu 0/1 pass, 0 running, 1 queued
beanstalk/windows+dgpu 0/1 pass, 1 error
beanstalk/linux 0/1 pass, 1 error
2026-11-16 13:28:29 +00:00
Compare
Author
First-time contributor

Rebuilt perf/apu-sleep-timer cleanly onto bbe421f (842dedb test failing on main -> 3dde802 fix passing) and addressed all four points from @dyst's review:

  1. Reach vs evidence: The branch now uses Option B (HOS_OPT_INTEGRATED_GPU = 62 in gpu/vk/vk_api.h and gpu/vk/vk_replay.cpp), scoping the unset HOS_SHORT_SLEEP_TIMER how = 1 default on Windows strictly to VK_PHYSICAL_DEVICE_TYPE_INTEGRATED_GPU. Windows discrete GPUs (> 6 logical processors) remain unchanged on main's processor threshold and HOS_SHORT_SLEEP_POLLERS cap. The PR description replaces the earlier single-run rc.5 numbers with a 2x2 interleaved compiled-binary A/B in ZoraDomain_Child (HOS_SHORT_SLEEP_TIMER unset, testing the shipped how = 1 path directly with identical settings on both arms): +15.3% Locked-View wall FPS (24.32 -> 28.05 fps), +13.8% settled log FPS (25.0 -> 28.45 fps), -22.3% GPU frame time (28.3 -> 22.0 ms), and > 40 ms present intervals down from 32–76 to 3–4.
  2. Serial threads under how = 1: In our v3.0.0-rc.5 single-run sweep (both stacked on RX_VK_PASS_BARRIER=game + RX_VK_READBACK_LATE=newest), HOS_SHORT_SLEEP_POLLERS=1 (how = 0, cap = 1, where serial threads stay exempt and yield) gave 33.39 Locked-View FPS vs 32.93 with HOS_SHORT_SLEEP_TIMER=1 (how = 2), but had higher GPU frame time (24.58 ms vs 21.35 ms) and lower tail average FPS (26.43 vs 28.73 fps). As noted in host/hos_thread.c:1194 (and 1204-1206), sead::MainThread itself yields 3.3–3.6 ms/frame while waiting on GPU readbacks; letting serial threads keep yielding on an integrated GPU leaves MainThread burning a core out of the shared APU package TDP during readbacks. Under how = 1, host short sleeps (hos_sleep_ns with guest = 0: frame limiter, audio renderer) still satisfy short_timer < 2 (host/hos_thread.c:1225) and continue to yield.
  3. Stale comments & docs: Restored main's const int cap = how || getenv("HOS_SHORT_SLEEP_TIMER") ? 0 : ... (host/hos_thread.c:1217), so explicit HOS_SHORT_SLEEP_TIMER=0 keeps cap = 0 (all yield) matching the log line. Updated the comments above hos_short_sleep_timer() (1104-1120), inside sleep_ns() (1181-1185), and in the Windows #else branch (1202-1206), and added the HOS_SHORT_SLEEP_TIMER=0|1, HOS_SHORT_SLEEP_POLLERS=<n> row to docs/development.md (### Running a test).
  4. test_short_sleep_default.py: Removed tests/test_short_sleep_default.py and folded the HOS_OPT_INTEGRATED_GPU = 62 wiring checks (gpu/vk/vk_api.h and gpu/vk/vk_replay.cpp t.option) directly into tests/test_short_sleep_timer.py alongside the C fixture's rx_option(62, 0) stub and igpu-ignored / dgpu-forced mutant checks (--cc gcc and --cc clang both pass).
Rebuilt `perf/apu-sleep-timer` cleanly onto `bbe421f` (`842dedb` test failing on `main` -> `3dde802` fix passing) and addressed all four points from @dyst's review: 1. **Reach vs evidence:** The branch now uses Option B (`HOS_OPT_INTEGRATED_GPU = 62` in `gpu/vk/vk_api.h` and `gpu/vk/vk_replay.cpp`), scoping the unset `HOS_SHORT_SLEEP_TIMER` `how = 1` default on Windows strictly to `VK_PHYSICAL_DEVICE_TYPE_INTEGRATED_GPU`. Windows discrete GPUs (`> 6` logical processors) remain unchanged on `main`'s processor threshold and `HOS_SHORT_SLEEP_POLLERS` cap. The PR description replaces the earlier single-run rc.5 numbers with a 2x2 interleaved compiled-binary A/B in `ZoraDomain_Child` (`HOS_SHORT_SLEEP_TIMER` unset, testing the shipped `how = 1` path directly with identical settings on both arms): `+15.3%` Locked-View wall FPS (`24.32 -> 28.05 fps`), `+13.8%` settled log FPS (`25.0 -> 28.45 fps`), `-22.3%` GPU frame time (`28.3 -> 22.0 ms`), and `> 40 ms` present intervals down from `32–76` to `3–4`. 2. **Serial threads under `how = 1`:** In our `v3.0.0-rc.5` single-run sweep (both stacked on `RX_VK_PASS_BARRIER=game` + `RX_VK_READBACK_LATE=newest`), `HOS_SHORT_SLEEP_POLLERS=1` (`how = 0, cap = 1`, where serial threads stay exempt and yield) gave `33.39` Locked-View FPS vs `32.93` with `HOS_SHORT_SLEEP_TIMER=1` (`how = 2`), but had higher GPU frame time (`24.58 ms` vs `21.35 ms`) and lower tail average FPS (`26.43` vs `28.73 fps`). As noted in `host/hos_thread.c:1194` (and `1204-1206`), `sead::MainThread` itself yields **3.3–3.6 ms/frame** while waiting on GPU readbacks; letting serial threads keep yielding on an integrated GPU leaves `MainThread` burning a core out of the shared APU package TDP during readbacks. Under `how = 1`, host short sleeps (`hos_sleep_ns` with `guest = 0`: frame limiter, audio renderer) still satisfy `short_timer < 2` (`host/hos_thread.c:1225`) and continue to yield. 3. **Stale comments & docs:** Restored `main`'s `const int cap = how || getenv("HOS_SHORT_SLEEP_TIMER") ? 0 : ...` (`host/hos_thread.c:1217`), so explicit `HOS_SHORT_SLEEP_TIMER=0` keeps `cap = 0` (`all yield`) matching the log line. Updated the comments above `hos_short_sleep_timer()` (`1104-1120`), inside `sleep_ns()` (`1181-1185`), and in the Windows `#else` branch (`1202-1206`), and added the `HOS_SHORT_SLEEP_TIMER=0|1`, `HOS_SHORT_SLEEP_POLLERS=<n>` row to `docs/development.md` (`### Running a test`). 4. **`test_short_sleep_default.py`:** Removed `tests/test_short_sleep_default.py` and folded the `HOS_OPT_INTEGRATED_GPU = 62` wiring checks (`gpu/vk/vk_api.h` and `gpu/vk/vk_replay.cpp` `t.option`) directly into `tests/test_short_sleep_timer.py` alongside the C fixture's `rx_option(62, 0)` stub and `igpu-ignored` / `dgpu-forced` mutant checks (`--cc gcc` and `--cc clang` both pass).
Owner

@Beanstalk

@Beanstalk
Member

Beanstalk test farm

Commit 3dde802 · tier reviewed · approved by HumanBean · dashboard

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

Windows integrated and discrete GPUs and a Linux PC will compare main and the PR in Zora’s Domain, using the reported Minimum preset. Two alternating runs per side take about 22 minutes per tester and measure frame times, FPS, GPU time and CPU use, with a scene screenshot. Crashes, device loss, major performance regressions or a significant visual difference fail the plan; a Windows integrated-GPU report may wait for a matching volunteer. No tester online now matches windows+igpu: that target waits for one.

  • Where: windows+igpu (1 report), windows+dgpu (1 report), linux (1 report).
  • zora-domain: from the repository's save kokiri-start (settings quality_preset=1, handheld_profile=1), then teleport to Zora's Domain, measure domain for 30 s (after 10 s warm-up), take the picture domain-view to compare with main; 2 runs on each side.
  • Measured: crashes, frame times, stutters and hitches, draws and frame rate, GPU time per pass, memory and CPU time, the game's log.
  • The PR fails if: it crashes; or frame rate is more than 10% below main; or a picture differs from main by more than 5%; or GPU time a frame is more than 20% above main; or frame time (p99) is more than 20% above main; or forbidden log lines is above 0.
  • Time: about 23 min of game runs per tester (4 game starts), plus the two builds.

Started.

Target Reports Crash Frame p99 (ms) vs main VRAM peak (MB) vs main Result
windows+igpu 0/1 — — — queued
windows+dgpu · 1: windows+nvidia+nvidia-proprietary+dgpu NVIDIA GeForce RTX 4070, 621.1.0 1/1 1 none — — error
linux · 1: linux+amd+radv+igpu+handheld AMD Custom GPU 0932 (RADV VANGOGH), 26.1.99 1/1 1 none — — error

Measurements · windows+dgpu

Verdict: could not be judged on windows+dgpu (1/1 report).

Hardware: 1: windows+nvidia+nvidia-proprietary+dgpu NVIDIA GeForce RTX 4070, 621.1.0 13th Gen Intel(R) Core(TM) i7-13700F, 32 GB RAM
Scene: zora-domain, from the repository's save kokiri-start (quality_preset=1, handheld_profile=1); 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); zora-domain: 2 game starts a side, domain 30 s after 10 s warm-up; each number the median of the runs, then of 1 report
How to reproduce: the farm's plan above (@Beanstalk on this PR runs it again)

Before: main @ bbe421f After: this PR @ 3dde802
zora-domain · forbidden log lines 0.0 0.0

Flags and settings

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

Markdown for the PR description
## Measurements

**Verdict:** could not be judged on windows+dgpu (1/1 report).

Hardware: 1: windows+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce RTX 4070, 621.1.0` `13th Gen Intel(R) Core(TM) i7-13700F, 32 GB RAM`  
Scene: zora-domain, from the repository's save kokiri-start (quality_preset=1, handheld_profile=1); 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); zora-domain: 2 game starts a side, domain 30 s after 10 s warm-up; each number the median of the runs, then of 1 report  
How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again)

| | Before: main @ bbe421f | After: this PR @ 3dde802 |
|---|---|---|
| zora-domain · forbidden log lines | 0.0 | 0.0 |

### Flags and settings

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

Measurements · linux

Verdict: could not be judged on linux (1/1 report).

Hardware: 1: linux+amd+radv+igpu+handheld AMD Custom GPU 0932 (RADV VANGOGH), 26.1.99 AMD Custom APU 0932, 16 GB RAM
Scene: zora-domain, from the repository's save kokiri-start (quality_preset=1, handheld_profile=1); 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); zora-domain: 2 game starts a side, domain 30 s after 10 s warm-up; each number the median of the runs, then of 1 report
How to reproduce: the farm's plan above (@Beanstalk on this PR runs it again)

Before: main @ bbe421f After: this PR @ 3dde802
zora-domain · forbidden log lines 0.0 0.0

Flags and settings

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

Markdown for the PR description
## Measurements

**Verdict:** could not be judged on linux (1/1 report).

Hardware: 1: linux+amd+radv+igpu+handheld `AMD Custom GPU 0932 (RADV VANGOGH), 26.1.99` `AMD Custom APU 0932, 16 GB RAM`  
Scene: zora-domain, from the repository's save kokiri-start (quality_preset=1, handheld_profile=1); 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); zora-domain: 2 game starts a side, domain 30 s after 10 s warm-up; each number the median of the runs, then of 1 report  
How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again)

| | Before: main @ bbe421f | After: this PR @ 3dde802 |
|---|---|---|
| zora-domain · forbidden log lines | 0.0 | 0.0 |

### Flags and settings

None: the plan varied no setting, switch or build option.
<!-- beanstalk:sticky --> ### Beanstalk test farm Commit `3dde802` · tier **reviewed** · approved by HumanBean · [dashboard](https://beanstalk.magicbean.dev/prs/1114) **What will be tested** · AI plan (`gpt-6.1-terra`) Windows integrated and discrete GPUs and a Linux PC will compare main and the PR in Zora’s Domain, using the reported Minimum preset. Two alternating runs per side take about 22 minutes per tester and measure frame times, FPS, GPU time and CPU use, with a scene screenshot. Crashes, device loss, major performance regressions or a significant visual difference fail the plan; a Windows integrated-GPU report may wait for a matching volunteer. No tester online now matches windows+igpu: that target waits for one. - **Where:** windows+igpu (1 report), windows+dgpu (1 report), linux (1 report). - **zora-domain:** from the repository's save `kokiri-start` (settings quality_preset=1, handheld_profile=1), then teleport to Zora's Domain, measure domain for 30 s (after 10 s warm-up), take the picture domain-view to compare with main; 2 runs on each side. - **Measured:** crashes, frame times, stutters and hitches, draws and frame rate, GPU time per pass, memory and CPU time, the game's log. - **The PR fails if:** it crashes; or frame rate is more than 10% below main; or a picture differs from main by more than 5%; or GPU time a frame is more than 20% above main; or frame time (p99) is more than 20% above main; or forbidden log lines is above 0. - **Time:** about 23 min of game runs per tester (4 game starts), plus the two builds. Started. | Target | Reports | Crash | Frame p99 (ms) vs main | VRAM peak (MB) vs main | Result | | --- | --- | --- | --- | --- | --- | | windows+igpu | 0/1 | — | — | — | queued | | windows+dgpu · 1: windows+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce RTX 4070, 621.1.0` | 1/1 [1](https://beanstalk.magicbean.dev/reports/rp_94d2a99b63f206c2) | none | — | — | error | | linux · 1: linux+amd+radv+igpu+handheld `AMD Custom GPU 0932 (RADV VANGOGH), 26.1.99` | 1/1 [1](https://beanstalk.magicbean.dev/reports/rp_1d1ca340d4341522) | none | — | — | error | #### Measurements · windows+dgpu **Verdict:** could not be judged on windows+dgpu (1/1 report). Hardware: 1: windows+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce RTX 4070, 621.1.0` `13th Gen Intel(R) Core(TM) i7-13700F, 32 GB RAM` Scene: zora-domain, from the repository's save kokiri-start (quality_preset=1, handheld_profile=1); 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); zora-domain: 2 game starts a side, domain 30 s after 10 s warm-up; each number the median of the runs, then of 1 report How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again) | | Before: main @ bbe421f | After: this PR @ 3dde802 | |---|---|---| | zora-domain · forbidden log lines | 0.0 | 0.0 | ### Flags and settings None: the plan varied no setting, switch or build option. <details><summary>Markdown for the PR description</summary> ```markdown ## Measurements **Verdict:** could not be judged on windows+dgpu (1/1 report). Hardware: 1: windows+nvidia+nvidia-proprietary+dgpu `NVIDIA GeForce RTX 4070, 621.1.0` `13th Gen Intel(R) Core(TM) i7-13700F, 32 GB RAM` Scene: zora-domain, from the repository's save kokiri-start (quality_preset=1, handheld_profile=1); 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); zora-domain: 2 game starts a side, domain 30 s after 10 s warm-up; each number the median of the runs, then of 1 report How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again) | | Before: main @ bbe421f | After: this PR @ 3dde802 | |---|---|---| | zora-domain · forbidden log lines | 0.0 | 0.0 | ### Flags and settings None: the plan varied no setting, switch or build option. ``` </details> #### Measurements · linux **Verdict:** could not be judged on linux (1/1 report). Hardware: 1: linux+amd+radv+igpu+handheld `AMD Custom GPU 0932 (RADV VANGOGH), 26.1.99` `AMD Custom APU 0932, 16 GB RAM` Scene: zora-domain, from the repository's save kokiri-start (quality_preset=1, handheld_profile=1); 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); zora-domain: 2 game starts a side, domain 30 s after 10 s warm-up; each number the median of the runs, then of 1 report How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again) | | Before: main @ bbe421f | After: this PR @ 3dde802 | |---|---|---| | zora-domain · forbidden log lines | 0.0 | 0.0 | ### Flags and settings None: the plan varied no setting, switch or build option. <details><summary>Markdown for the PR description</summary> ```markdown ## Measurements **Verdict:** could not be judged on linux (1/1 report). Hardware: 1: linux+amd+radv+igpu+handheld `AMD Custom GPU 0932 (RADV VANGOGH), 26.1.99` `AMD Custom APU 0932, 16 GB RAM` Scene: zora-domain, from the repository's save kokiri-start (quality_preset=1, handheld_profile=1); 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); zora-domain: 2 game starts a side, domain 30 s after 10 s warm-up; each number the median of the runs, then of 1 report How to reproduce: the farm's plan above (`@Beanstalk` on this PR runs it again) | | Before: main @ bbe421f | After: this PR @ 3dde802 | |---|---|---| | zora-domain · forbidden log lines | 0.0 | 0.0 | ### Flags and settings None: the plan varied no setting, switch or build option. ``` </details>
futondestroyer left a comment

Code-and-evidence re-review of exact head 3dde802f4770648d1af92db52c757b0c4e4ead89. No tests, builds or gameplay run for this review.

No blocking code finding. The current branch addresses the scope and evidence concerns in dyst's earlier review: the extra unset-default override is Windows-only and conditional on the selected Vulkan device being integrated. Explicit HOS_SHORT_SLEEP_TIMER values retain precedence, discrete GPUs retain the processor/poller rule, and the Linux branch keeps its existing behavior. The existing option callback carries the new query without changing the callback structures. Host short sleeps continue through the short_timer < (guest ? 1 : 2) distinction; this is not the former blanket timer-for-all-sleeps proposal.

The changed test compiles the real sleep path with timer/yield/option stand-ins, retains discrete-GPU and explicit-setting cases, and adds integrated-GPU cases and mutations that reject ignoring detection or forcing the discrete path onto the timer. That is materially stronger evidence than the superseded copied Python decision. It does not measure real timer latency, package power, driver classification or startup on another machine. The reported APU evidence now uses two interleaved pairs of the shipped unset-default path; it supports the claim on that measured setup, not every Windows iGPU.

Remaining readiness: the description's required Windows discrete-GPU and Linux runs have not reported results yet, and current CI checks are pending. HumanBean already authorized this exact head on the farm, which currently has Windows discrete-GPU and Linux jobs running and a Windows iGPU job queued. No duplicate request or restart is needed. Keep HOS_SHORT_SLEEP_TIMER unset in those comparisons and verify the actual selected GPU and sleep-policy log, especially on hybrid systems.

Verdict: code looks ready; merge approval deferred until the required non-regression results and CI are available. This comment does not claim a personal hardware pass.

Code-and-evidence re-review of exact head `3dde802f4770648d1af92db52c757b0c4e4ead89`. No tests, builds or gameplay run for this review. No blocking code finding. The current branch addresses the scope and evidence concerns in dyst's earlier review: the extra unset-default override is Windows-only and conditional on the selected Vulkan device being integrated. Explicit HOS_SHORT_SLEEP_TIMER values retain precedence, discrete GPUs retain the processor/poller rule, and the Linux branch keeps its existing behavior. The existing option callback carries the new query without changing the callback structures. Host short sleeps continue through the `short_timer < (guest ? 1 : 2)` distinction; this is not the former blanket timer-for-all-sleeps proposal. The changed test compiles the real sleep path with timer/yield/option stand-ins, retains discrete-GPU and explicit-setting cases, and adds integrated-GPU cases and mutations that reject ignoring detection or forcing the discrete path onto the timer. That is materially stronger evidence than the superseded copied Python decision. It does not measure real timer latency, package power, driver classification or startup on another machine. The reported APU evidence now uses two interleaved pairs of the shipped unset-default path; it supports the claim on that measured setup, not every Windows iGPU. Remaining readiness: the description's required Windows discrete-GPU and Linux runs have not reported results yet, and current CI checks are pending. HumanBean already authorized this exact head on the farm, which currently has Windows discrete-GPU and Linux jobs running and a Windows iGPU job queued. No duplicate request or restart is needed. Keep HOS_SHORT_SLEEP_TIMER unset in those comparisons and verify the actual selected GPU and sleep-policy log, especially on hybrid systems. Verdict: code looks ready; merge approval deferred until the required non-regression results and CI are available. This comment does not claim a personal hardware pass.
Some checks reported errors
CI / Conventional Commit title (pull_request) Blocked by required conditions
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Blocked by required conditions
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Blocked by required conditions
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Blocked by required conditions
Guide screenshots / Guide images and local UI source review (pull_request) Blocked by required conditions
Third-party parts / As their notes say (pull_request) Blocked by required conditions
Third-party parts / The same as upstream (pull_request) Blocked by required conditions
CI / Workflow lint (actionlint) (pull_request) Blocked by required conditions
Website / Website on Cloudflare (pull_request) Blocked by required conditions
PR description / Description follows the template (pull_request_target) Successful in 4s
Website / Website on Cloudflare (pull_request_target) Has been skipped
beanstalk/approval approved by HumanBean (reviewed)
beanstalk/windows+igpu 0/1 pass, 0 running, 1 queued
beanstalk/windows+dgpu 0/1 pass, 1 error
beanstalk/linux 0/1 pass, 1 error
This pull request has changes conflicting with the target branch.
  • gpu/vk/vk_api.h
  • gpu/vk/vk_replay.cpp
Some workflows are waiting to be reviewed.
View command line instructions

Manual merge helper

Use this merge commit message when completing the merge manually.

Checkout

From your project repository, check out a new branch and test the changes.
git fetch -u 'perf/apu-sleep-timer:Jer9-perf/apu-sleep-timer'
git switch 'Jer9-perf/apu-sleep-timer'
Sign in to join this conversation.
No reviewers
No milestone
No project
No assignees
5 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!1114
No description provided.