ShuriZma
/
xenia-canary
mirror of https://github.com/xenia-canary/xenia-canary.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

[D3D12] RT dump: Fix r# allocation

This commit is contained in:
Triang3l 2022-03-22 21:41:44 +03:00
parent 920704c71a
commit ee8e71cea8
1 changed files with 2 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
src/xenia/gpu/d3d12/d3d12_render_target_cache.cc
View File

@ -5970,7 +5970,8 @@ ID3D12PipelineState* D3D12RenderTargetCache::GetOrCreateDumpPipeline(
a.OpDclInput(dxbc::Dest::VThreadID(0b0011)); a.OpDclInput(dxbc::Dest::VThreadID(0b0011));
// r0 - addressing before the load, then addressing and conversion scratch // r0 - addressing before the load, then addressing and conversion scratch
// r1 - addressing scratch before the load, then data // r1 - addressing scratch before the load, then data
a.OpDclTemps(2); stat.temp_register_count = 2;
a.OpDclTemps(stat.temp_register_count);
// There's no strict dependency on the group size here, for simplicity of // There's no strict dependency on the group size here, for simplicity of
// calculations especially with resolution scaling, dividing manually (as the // calculations especially with resolution scaling, dividing manually (as the
// group size is not unlimited). The only restriction is that an integer // group size is not unlimited). The only restriction is that an integer
@ -6355,8 +6356,6 @@ ID3D12PipelineState* D3D12RenderTargetCache::GetOrCreateDumpPipeline(
case xenos::ColorRenderTargetFormat::k_2_10_10_10_FLOAT_AS_16_16_16_16: case xenos::ColorRenderTargetFormat::k_2_10_10_10_FLOAT_AS_16_16_16_16:
// Float16 has a wider range for both color and alpha, also NaNs. // Float16 has a wider range for both color and alpha, also NaNs.
// Color - clamp and convert. // Color - clamp and convert.
stat.temp_register_count =
std::max(stat.temp_register_count, uint32_t(3));
// Convert red in r1.x to the result register r1.x - the same, but // Convert red in r1.x to the result register r1.x - the same, but
// UnclampedFloat32To7e3 allows that - using r0.x as a temporary. // UnclampedFloat32To7e3 allows that - using r0.x as a temporary.
DxbcShaderTranslator::UnclampedFloat32To7e3(a, 1, 0, 1, 0, 0, 0); DxbcShaderTranslator::UnclampedFloat32To7e3(a, 1, 0, 1, 0, 0, 0);