Run clang-format (the repo's .clang-format: LLVM base, 120 cols, tabs,
Allman functions) over every source file so the tree is formatter-clean.
Whitespace only -- no behavior change; full x64 + x86 suites pass.
Also set SortIncludes: false in .clang-format. Windows include order is
load-bearing (windows.h must precede tlhelp32.h / mmreg.h / xinput.h /
dinput.h; winsock2.h must precede windows.h), and the default
alphabetical sort reorders tlhelp32.h ahead of windows.h -- a build
break. Leaving order alone keeps the manual, correct grouping.
Against Sphere Spectacle (144 FPS, runs without Steam) the implicit-layer
capture dropped the game to ~3 FPS. Measured cause (per-stage trace in the
layer): the read-back ran on the game's PRESENT THREAD and spent ~370 ms per
1080p frame -- not the GPU copy (~2 ms) but the CPU swizzle, because the staging
buffer was a plain HOST_VISIBLE|HOST_COHERENT type (write-combined / uncached on
a discrete GPU), where a scattered CPU read runs at PCIe latency. 3 captures/s =
the 3 FPS the user saw.
Test-first: tests/vk_capture_perf_test reproduces the stall as a deterministic
unit test (372 ms/present, ratio 1.0 -> FAIL via `--sync`), then proves the fix
(0.02 ms/present, byte-correct BGRA->RGBA, ratio ~0 -> PASS).
Fix: extract the near-identical read-back from vk_hook.cpp and coop_vk_layer.cpp
into one shared coop::hook::VkCapture that:
* has the present thread only record + submit the copy (sub-ms) and return;
* runs a dedicated reaper thread for the fence wait + swizzle + D3D upload, off
the critical path, with a ring of in-flight slots (game never waits);
* allocates HOST_CACHED staging (fast CPU read), invalidating when non-coherent;
* throttles capture to ~150 Hz (a guest stream is <= the host refresh; no point
mirroring an uncapped 400+ FPS game and burning reaper CPU).
Real-game A/B: present rate now matches the no-capture baseline (605->470 vs
593->405 over the same ramp) with the mirror at ~130 fps -- no measurable impact.
Also adds present-thread overhead guards to the other GPU backends' hook tests
(present_overhead.hpp): DX11 0.05 ms, DX12 0.34 ms, OpenGL 0.09 ms overhead, all
asserted < one 60 Hz frame, so any future synchronous-stall regression fails.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
D3D12 games (e.g. Spider-Man: Miles Morales) fire the Present hook -- the DXGI
swapchain's Present is the same vtable function for D3D11 and D3D12 -- but
GetBuffer(0) as ID3D11Texture2D fails, so the hook used to idle. Now, when the D3D11
GetBuffer fails, present_hook bridges via D3D11On12: it gets the game's ID3D12Device
from the backbuffer, creates its own DIRECT command queue on it (no need to hook the
game's ExecuteCommandLists), builds an ID3D11On12Device, CreateWrappedResource's the
D3D12 backbuffer, and CopyResource's it into the existing shared keyed-mutex texture
-- so the host side is unchanged. The bridge is created lazily and torn down with the
hook.
Verified with a new in-process dx12_present_hook_test: it drives a real D3D12
swapchain (clears a backbuffer, Presents) and asserts present fired, the backbuffer
was bridged into the shared texture, and a second device reads the exact color back
by name -- {51,102,153,255}. Full build x64 + x86 clean (the x86 hook compiles the
D3D12 path too); ctest x64 10/10, x86 3/3.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
