moonlight-stream/moonlight-qt
1
0
Fork 0
mirror of https://github.com/moonlight-stream/moonlight-qt.git synced 2026-10-11 19:13:59 +00:00
Code Issues Packages Projects Releases Wiki Activity

Allow separate devices for Vega+ and block them for Fermi

Browse Source 
Fermi is known broken, Kepler and Maxwell v1 may work but are currently untested
This commit is contained in:
Cameron Gutman
2026-09-06 16:12:52 -05:00
parent 09873287ca
commit b86b5f71e3
1 changed files with 15 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files
app/streaming/video/ffmpeg-renderers/d3d11va.cpp
+15 -7
View File
@@ -371,19 +371,28 @@ bool D3D11VARenderer::createDeviceByAdapterIndex(int adapterIndex, bool* adapter
separateDevices = SUCCEEDED(hr) && d3d11Options.ExtendedResourceSharing && m_FenceType != SupportedFenceType::None;
if (separateDevices) {
// Use minimum precision support to differentiate Vega and later from Polaris and earlier
D3D11_FEATURE_DATA_SHADER_MIN_PRECISION_SUPPORT minPrecSupport;
hr = m_RenderDevice->CheckFeatureSupport(D3D11_FEATURE_SHADER_MIN_PRECISION_SUPPORT, &minPrecSupport, sizeof(minPrecSupport));
if (FAILED(hr)) {
minPrecSupport = {};
}
// This texture array sharing codepath is quite prone to driver bugs.
//
// Broken GPU vendors/cards/drivers include:
// - Moore Threads (texture is all zero/green)
// - Qualcomm (decoding is unstable/slow on QC710)
// - AMD (Polaris cards display corrupt output - see #2003,
// HD 5570 drivers deadlock with shared texture arrays)
// - AMD prior to Vega (Polaris cards display corrupt output - see #2003,
// HD 5570 drivers deadlock with shared texture arrays)
// - Nvidia prior to Maxwell 2? (Fermi cards display all zero/green)
//
// Due to all these issues, we will only use this path for Intel and NVIDIA where we know it
// Due to all these issues, we will only use this path for Intel/AMD and NVIDIA where we know it
// provides tangible benefits (performance for the former and VRR support for the latter).
if (adapterDesc.VendorId != 0x8086 && adapterDesc.VendorId != 0x10DE && adapterDesc.VendorId != 'ADVN') {
separateDevices = false;
}
separateDevices = adapterDesc.VendorId == 0x8086 || // Intel
(adapterDesc.VendorId == 0x10DE && featureLevel >= D3D_FEATURE_LEVEL_11_1) || // NVIDIA Maxwell 2+ (PCI ID)
adapterDesc.VendorId == 'ADVN' || // NVIDIA (WoA)
(adapterDesc.VendorId == 0x1002 && minPrecSupport.PixelShaderMinPrecision == D3D11_SHADER_MIN_PRECISION_16_BIT); // AMD Vega+
}
}
@@ -406,7 +415,6 @@ bool D3D11VARenderer::createDeviceByAdapterIndex(int adapterIndex, bool* adapter
// Also bind SRVs as this improves render times by about 2x on
// my Ryzen 3300U system.
m_BindDecoderOutputTextures = adapterDesc.VendorId == 0x8086 ||
(adapterDesc.VendorId == 0x1002 && featureLevel >= D3D_FEATURE_LEVEL_11_1) ||
separateDevices;
}