moonlight-stream/moonlight-qt
1
0
Fork 0
mirror of https://github.com/moonlight-stream/moonlight-qt.git synced 2025-07-01 07:15:27 +00:00
Code Issues Packages Projects Releases Wiki Activity

Don't use the bind path for UMA GPUs

Browse Source 
It turns out to be very marginally slower on RDNA2 iGPUs.
This commit is contained in:
Cameron Gutman 2024-09-12 21:32:36 -05:00
parent 6ec0c79899
commit 92ee4a3046
1 changed files with 1 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
app/streaming/video/ffmpeg-renderers/d3d11va.cpp
View File

@ -233,22 +233,10 @@ bool D3D11VARenderer::createDeviceByAdapterIndex(int adapterIndex, bool* adapter
bool ok;
m_BindDecoderOutputTextures = !!qEnvironmentVariableIntValue("D3D11VA_FORCE_BIND", &ok);
if (!ok) {
D3D11_FEATURE_DATA_D3D11_OPTIONS2 options = {};
m_Device->CheckFeatureSupport(D3D11_FEATURE_D3D11_OPTIONS2, &options, sizeof(options));
SDL_LogInfo(SDL_LOG_CATEGORY_APPLICATION,
"Unified memory: %s",
options.UnifiedMemoryArchitecture ? "yes" : "no");
// Skip copying to our own internal texture on Intel GPUs due to
// significant performance impact of the extra copy. See:
// https://github.com/moonlight-stream/moonlight-qt/issues/1304
//
// We also don't copy for modern UMA GPUs from other vendors to
// avoid performance impact due to shared system memory accesses.
m_BindDecoderOutputTextures =
adapterDesc.VendorId == 0x8086 ||
(featureLevel >= D3D_FEATURE_LEVEL_11_1 && options.UnifiedMemoryArchitecture);
m_BindDecoderOutputTextures = adapterDesc.VendorId == 0x8086;
}
else {
SDL_LogWarn(SDL_LOG_CATEGORY_APPLICATION,