piwalker/dolphin
1
0
Fork 0
mirror of https://github.com/dolphin-emu/dolphin.git synced 2025-07-29 00:59:44 -06:00
Code Issues Packages Projects Releases Wiki Activity

Added SSSE3/SSE4.1 code for speed up. The code does not work in this revision because cpu_info is not initialized properly. I will fix the issue in another commit.

Browse Source 
git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@5119 8ced0084-cf51-0410-be5f-012b33b47a6e
This commit is contained in:
nodchip
2010-02-24 09:56:41 +00:00
parent dc9d533d4f
commit 47fb73b71a
4 changed files with 189 additions and 30 deletions
Download Patch File Download Diff File Expand all files Collapse all files

32
Source/Core/VideoCommon/Src/VertexLoader.cpp
View File

@ -16,8 +16,10 @@
// http://code.google.com/p/dolphin-emu/
#include <assert.h>
#include <nmmintrin.h>
#include "Common.h"
#include "CPUDetect.h"
#include "VideoCommon.h"
#include "VideoConfig.h"
#include "Profiler.h"
@ -70,7 +72,7 @@ int colIndex;
TVtxAttr* pVtxAttr;
int colElements[2];
float posScale;
float tcScale[8];
__declspec(align(16)) float tcScale[8];
using namespace Gen;
@ -633,9 +635,31 @@ void VertexLoader::RunVertices(int vtx_attr_group, int primitive, int count)
pVtxAttr = &m_VtxAttr;
posScale = 1.0f / float(1 << m_VtxAttr.PosFrac);
if (m_NativeFmt->m_components & VB_HAS_UVALL)
for (int i = 0; i < 8; i++)
tcScale[i] = 1.0f / float(1 << m_VtxAttr.texCoord[i].Frac);
if (m_NativeFmt->m_components & VB_HAS_UVALL) {
if (cpu_info.bSSE4_1) {
__m128i a0;
a0 = _mm_insert_epi32(a0, 1 << m_VtxAttr.texCoord[0].Frac, 0);
a0 = _mm_insert_epi32(a0, 1 << m_VtxAttr.texCoord[1].Frac, 1);
a0 = _mm_insert_epi32(a0, 1 << m_VtxAttr.texCoord[2].Frac, 2);
a0 = _mm_insert_epi32(a0, 1 << m_VtxAttr.texCoord[3].Frac, 3);
const __m128 b0 = _mm_cvtepi32_ps(a0);
const __m128 c0 = _mm_rcp_ps(b0);
_mm_stream_ps(&tcScale[0], c0);
__m128i a1;
a1 = _mm_insert_epi32(a1, 1 << m_VtxAttr.texCoord[4].Frac, 0);
a1 = _mm_insert_epi32(a1, 1 << m_VtxAttr.texCoord[5].Frac, 1);
a1 = _mm_insert_epi32(a1, 1 << m_VtxAttr.texCoord[6].Frac, 2);
a1 = _mm_insert_epi32(a1, 1 << m_VtxAttr.texCoord[7].Frac, 3);
const __m128 b1 = _mm_cvtepi32_ps(a1);
const __m128 c1 = _mm_rcp_ps(b1);
_mm_stream_ps(&tcScale[4], c1);
} else {
for (int i = 0; i < 8; i++) {
tcScale[i] = 1.0f / float(1 << m_VtxAttr.texCoord[i].Frac);
}
}
}
for (int i = 0; i < 2; i++)
colElements[i] = m_VtxAttr.color[i].Elements;