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

Vertex loader: optimize texmtx_write_float4 

Seems to be pretty high in the profile in some geometry-heavy games like The
Last Story, and the compiler-generated assembly is terrifyingly bad, so
SSE-ize it.
This commit is contained in:
Fiora 2014-11-30 10:50:36 -08:00
parent b9b3277fb5
commit e3578683e3
1 changed files with 6 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
Source/Core/VideoCommon/VertexLoader.cpp
View File

@ -91,11 +91,17 @@ static void LOADERDECL TexMtx_Write_Float2()
static void LOADERDECL TexMtx_Write_Float4() static void LOADERDECL TexMtx_Write_Float4()
{ {
#if _M_SSE >= 0x200
__m128 output = _mm_cvtsi32_ss(_mm_castsi128_ps(_mm_setzero_si128()), s_curtexmtx[s_texmtxwrite++]);
_mm_storeu_ps((float*)VertexManager::s_pCurBufferPointer, _mm_shuffle_ps(output, output, 0x45 /* 1, 1, 0, 1 */));
VertexManager::s_pCurBufferPointer += sizeof(float) * 4;
#else
DataWrite(0.f); DataWrite(0.f);
DataWrite(0.f); DataWrite(0.f);
DataWrite(float(s_curtexmtx[s_texmtxwrite++])); DataWrite(float(s_curtexmtx[s_texmtxwrite++]));
// Just to fill out with 0. // Just to fill out with 0.
DataWrite(0.f); DataWrite(0.f);
#endif
} }
VertexLoader::VertexLoader(const TVtxDesc &vtx_desc, const VAT &vtx_attr) VertexLoader::VertexLoader(const TVtxDesc &vtx_desc, const VAT &vtx_attr)