Skip to content

feat(d3d11_service): per-stage weave timing under DXR_WEAVE_GPU_TIMING (2D-under-the-lens P0) #1824

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
dfattal merged 1 commit into from
Oct 5, 2026
Merged

feat(d3d11_service): per-stage weave timing under DXR_WEAVE_GPU_TIMING (2D-under-the-lens P0) #1824

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions docs/roadmap/control-panel-performance-settings.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -518,6 +518,7 @@ as `docs/specs/vendor/oem-android-platform-requirements.md` §R6.
| `DXR_FRAME_WITNESS` | `compositor/util/comp_frame_witness.h:62` | `getenv`, atomic cached | 0 (secs) | App | 3 | Windowed count of app weaves vs. repaints vs. presents vs. 3D weaves |
| `DXR_WEAVE_LATENCY_CSV` | `compositor/util/comp_weave_latency_win.h:82`; `vk_native/comp_vk_native_target.cpp:1860, 1878` | `getenv` | unset | App | 3 | Filename prefix for latency CSV output |
| `DXR_DP_FORWARD_HORIZON_TRACE` | `compositor/util/comp_weave_latency_win.h` (`note_horizon`, called from `predict_weave_to_scanout_ns` on the split + d3d11 + d3d12 weave paths) | `getenv`, cached | off | App | 3 | #206 horizon-jitter probe: one throttled WARN per ~5 s with min/mean/max of the forward horizon, its spread in refresh periods, the count of vblank-boundary flips (consecutive-weave deltas ≥ half a period), and (#1432) the closed loop — resolved weaves, wrong-slot calls (|predicted − realised| ≥ half a period) and mean |error|. Pure observer — changes nothing but logging |
| `DXR_WEAVE_GPU_TIMING` | `compositor/d3d11_service/comp_d3d11_service.cpp` (`weave_gpu_timing_enabled`); also read by the Leia plug-in's D3D11 DP (`leia_dp_timing_enabled`, alias `DXR_LEIA_DP_WEAVE_TIMING`) | `getenv`, `static` cached | off | Svc | 3 | Per-stage cost of the XR_DXR_weave present-owner path (`comp_d3d11_service_weave_submit`): one WARN `weave timing [...]` line per client per 5 s with GPU avg/max for the #1058 output clear, the v6 crop copy / batch SBS clear+blit, the DP weave (`set_overlay_2d` + `process_atlas`), the post-weave overlay fallback blit and the epilogue; CPU avg/max for the `render_mutex` + `immediate_ctx_mutex` waits, the input and overlay keyed-mutex acquires and the whole synchronous submit; and accepted/refused submits by path and by 2D-overlay route (DP / fallback / none / unchanged). Non-blocking timestamp queries, read frames late. The plug-in adds its own `Leia D3D11 DP timing` line (SR weave vs alpha gate GPU time, 2D-layer compose + strength). Set it in the **service's** environment. Pure observer |
| `DXR_WEAVE_PROBE` | `compositor/d3d11/comp_d3d11_compositor.cpp:2383` | `getenv`, cached | off | App | 4 | One-shot dump of the back buffer as the DP left it, pre-composite |
| `DXR_WEAVE_TAP` | `compositor/d3d12/comp_d3d12_compositor.cpp:4122` | `getenv`, cached | 0 | App | 4 | #727 dual-tap PNG dump around the composite. GPU flush + readback; costly |
| `DXR_PHASE_DEBUG` | `compositor/d3d12/comp_d3d12_compositor.cpp:767` | `getenv`, magic static | off | App | 4 | #740 corner-vs-centre phase seed dump |
Expand Down
Loading
Loading