ShuriZma
/
pcsx2
mirror of https://github.com/PCSX2/pcsx2.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

GS: AVX2 WriteBlock functions

This commit is contained in:
TellowKrinkle 2021-04-07 13:45:25 -05:00 committed by refractionpcsx2
parent 263e495561
commit d64e838b6d
2 changed files with 42 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
pcsx2/GS/GSBlock.cpp
View File

@ -19,10 +19,13 @@
CONSTINIT const GSVector4i GSBlock::m_r16mask(0, 1, 4, 5, 8, 9, 12, 13, 2, 3, 6, 7, 10, 11, 14, 15);
CONSTINIT const GSVector4i GSBlock::m_r8mask(0, 4, 2, 6, 8, 12, 10, 14, 1, 5, 3, 7, 9, 13, 11, 15);
CONSTINIT const GSVector4i GSBlock::m_r4mask(0, 8, 4, 12, 1, 9, 5, 13, 2, 10, 6, 14, 3, 11, 7, 15);
CONSTINIT const GSVector4i GSBlock::m_w4mask(0, 4, 8, 12, 2, 6, 10, 14, 1, 5, 9, 13, 3, 7, 11, 15);
CONSTINIT const GSVector4i GSBlock::m_palvec_mask(0, 4, 8, 12, 1, 5, 9, 13, 2, 6, 10, 14, 3, 7, 11, 15);
CONSTINIT const GSVector4i GSBlock::m_avx2_r8mask1(0, 4, 8, 12, 1, 5, 9, 13, 2, 6, 10, 14, 3, 7, 11, 15);
CONSTINIT const GSVector4i GSBlock::m_avx2_r8mask2(1, 5, 9, 13, 0, 4, 8, 12, 3, 7, 11, 15, 2, 6, 10, 14);
CONSTINIT const GSVector4i GSBlock::m_avx2_w8mask1(0, 4, 8, 12, 1, 5, 9, 13, 2, 6, 10, 14, 3, 7, 11, 15);
CONSTINIT const GSVector4i GSBlock::m_avx2_w8mask2(4, 0, 12, 8, 5, 1, 13, 9, 6, 2, 14, 10, 7, 3, 15, 11);
CONSTINIT const GSVector4i GSBlock::m_uw8hmask0(0, 0, 0, 0, 1, 1, 1, 1, 8, 8, 8, 8, 9, 9, 9, 9);
CONSTINIT const GSVector4i GSBlock::m_uw8hmask1(2, 2, 2, 2, 3, 3, 3, 3, 10, 10, 10, 10, 11, 11, 11, 11);

50
pcsx2/GS/GSBlock.h
View File

@ -24,10 +24,13 @@ class GSBlock
static const GSVector4i m_r16mask;
static const GSVector4i m_r8mask;
static const GSVector4i m_r4mask;
static const GSVector4i m_w4mask;
static const GSVector4i m_palvec_mask;
static const GSVector4i m_avx2_r8mask1;
static const GSVector4i m_avx2_r8mask2;
static const GSVector4i m_avx2_w8mask1;
static const GSVector4i m_avx2_w8mask2;
static const GSVector4i m_uw8hmask0;
static const GSVector4i m_uw8hmask1;
@ -216,24 +219,20 @@ public:
GSVector8i v0(v4, v5);
GSVector8i v1(v6, v7);
GSVector8i v2 = v0.blend32<0xaa>(v1).shuffle8(GSVector8i::broadcast128(m_avx2_w8mask1)).acbd();
GSVector8i v3 = v1.blend32<0xaa>(v0).shuffle8(GSVector8i::broadcast128(m_avx2_w8mask2)).acbd();
if ((i & 1) == 0)
{
v1 = v1.yxwz();
((GSVector8i*)dst)[i * 2 + 0] = v2;
((GSVector8i*)dst)[i * 2 + 1] = v3;
}
else
{
v0 = v0.yxwz();
((GSVector8i*)dst)[i * 2 + 0] = v3;
((GSVector8i*)dst)[i * 2 + 1] = v2;
}
GSVector8i::sw8(v0, v1);
GSVector8i::sw16(v0, v1);
v0 = v0.acbd();
v1 = v1.acbd();
((GSVector8i*)dst)[i * 2 + 0] = v0;
((GSVector8i*)dst)[i * 2 + 1] = v1;
#else
GSVector4i v0 = GSVector4i::load<alignment != 0>(&src[srcpitch * 0]);
@ -273,6 +272,33 @@ public:
// TODO: pshufb
#if _M_SSE >= 0x501
GSVector8i v0 = GSVector8i(GSVector4i::load<false>(&src[srcpitch * 0]), GSVector4i::load<false>(&src[srcpitch * 1]));
GSVector8i v1 = GSVector8i(GSVector4i::load<false>(&src[srcpitch * 2]), GSVector4i::load<false>(&src[srcpitch * 3]));
v0 = v0.shuffle8(GSVector8i::broadcast128(m_w4mask)).acbd();
v1 = v1.shuffle8(GSVector8i::broadcast128(m_w4mask)).acbd();
if ((i & 1) == 0)
{
v0 = v0.xzyw();
v1 = v1.ywxz();
}
else
{
v0 = v0.ywxz();
v1 = v1.xzyw();
}
GSVector8i::mix4(v0, v1);
GSVector8i::sw32(v0, v1);
((GSVector8i*)dst)[i * 2 + 0] = v0;
((GSVector8i*)dst)[i * 2 + 1] = v1;
#else
GSVector4i v0 = GSVector4i::load<alignment != 0>(&src[srcpitch * 0]);
GSVector4i v1 = GSVector4i::load<alignment != 0>(&src[srcpitch * 1]);
GSVector4i v2 = GSVector4i::load<alignment != 0>(&src[srcpitch * 2]);
@ -298,6 +324,8 @@ public:
((GSVector4i*)dst)[i * 4 + 1] = v1;
((GSVector4i*)dst)[i * 4 + 2] = v2;
((GSVector4i*)dst)[i * 4 + 3] = v3;
#endif
}
template <int alignment, u32 mask>