mirror of
https://github.com/PCSX2/pcsx2.git
synced 2026-09-10 23:02:59 +02:00
Mostly code cleanups, XBYAK 2.99, VEX conversion for the sw renderer (3-5% faster), GSState::Move fix for dark cloud 2 invention crash.
git-svn-id: http://pcsx2.googlecode.com/svn/trunk@4287 96395faa-99c1-11dd-bbfe-3dabce05a288
This commit is contained in:
gabest11
@@ -48,7 +48,14 @@ void GSSetupPrimCodeGenerator::Generate()
|
||||
{
|
||||
for(int i = 0; i < 5; i++)
|
||||
{
|
||||
movaps(Xmm(3 + i), xmmword[&m_shift[i]]);
|
||||
if(m_cpu.has(util::Cpu::tAVX))
|
||||
{
|
||||
vmovaps(Xmm(3 + i), ptr[&m_shift[i]]);
|
||||
}
|
||||
else
|
||||
{
|
||||
movaps(Xmm(3 + i), ptr[&m_shift[i]]);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
@@ -68,113 +75,221 @@ void GSSetupPrimCodeGenerator::Depth()
|
||||
return;
|
||||
}
|
||||
|
||||
if(!m_sel.sprite)
|
||||
if(m_cpu.has(util::Cpu::tAVX))
|
||||
{
|
||||
// GSVector4 t = dscan.p;
|
||||
|
||||
movaps(xmm0, xmmword[edx + 16]);
|
||||
|
||||
if(m_en.f)
|
||||
if(!m_sel.sprite)
|
||||
{
|
||||
// GSVector4 df = p.wwww();
|
||||
// GSVector4 t = dscan.p;
|
||||
|
||||
movaps(xmm1, xmm0);
|
||||
shufps(xmm1, xmm1, _MM_SHUFFLE(3, 3, 3, 3));
|
||||
vmovaps(xmm0, ptr[edx + 16]);
|
||||
|
||||
// m_env.d4.f = GSVector4i(df * 4.0f).xxzzlh();
|
||||
|
||||
movaps(xmm2, xmm1);
|
||||
mulps(xmm2, xmm3);
|
||||
cvttps2dq(xmm2, xmm2);
|
||||
pshuflw(xmm2, xmm2, _MM_SHUFFLE(2, 2, 0, 0));
|
||||
pshufhw(xmm2, xmm2, _MM_SHUFFLE(2, 2, 0, 0));
|
||||
movdqa(xmmword[&m_env.d4.f], xmm2);
|
||||
|
||||
for(int i = 0; i < 4; i++)
|
||||
if(m_en.f)
|
||||
{
|
||||
// m_env.d[i].f = GSVector4i(df * m_shift[i]).xxzzlh();
|
||||
// GSVector4 df = p.wwww();
|
||||
|
||||
movaps(xmm2, xmm1);
|
||||
mulps(xmm2, Xmm(4 + i));
|
||||
cvttps2dq(xmm2, xmm2);
|
||||
pshuflw(xmm2, xmm2, _MM_SHUFFLE(2, 2, 0, 0));
|
||||
pshufhw(xmm2, xmm2, _MM_SHUFFLE(2, 2, 0, 0));
|
||||
movdqa(xmmword[&m_env.d[i].f], xmm2);
|
||||
vshufps(xmm1, xmm0, xmm0, _MM_SHUFFLE(3, 3, 3, 3));
|
||||
|
||||
// m_env.d4.f = GSVector4i(df * 4.0f).xxzzlh();
|
||||
|
||||
vmulps(xmm2, xmm1, xmm3);
|
||||
vcvttps2dq(xmm2, xmm2);
|
||||
vpshuflw(xmm2, xmm2, _MM_SHUFFLE(2, 2, 0, 0));
|
||||
vpshufhw(xmm2, xmm2, _MM_SHUFFLE(2, 2, 0, 0));
|
||||
vmovdqa(ptr[&m_env.d4.f], xmm2);
|
||||
|
||||
for(int i = 0; i < 4; i++)
|
||||
{
|
||||
// m_env.d[i].f = GSVector4i(df * m_shift[i]).xxzzlh();
|
||||
|
||||
vmulps(xmm2, xmm1, Xmm(4 + i));
|
||||
vcvttps2dq(xmm2, xmm2);
|
||||
vpshuflw(xmm2, xmm2, _MM_SHUFFLE(2, 2, 0, 0));
|
||||
vpshufhw(xmm2, xmm2, _MM_SHUFFLE(2, 2, 0, 0));
|
||||
vmovdqa(ptr[&m_env.d[i].f], xmm2);
|
||||
}
|
||||
}
|
||||
|
||||
if(m_en.z)
|
||||
{
|
||||
// GSVector4 dz = p.zzzz();
|
||||
|
||||
vshufps(xmm0, xmm0, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
|
||||
// m_env.d4.z = dz * 4.0f;
|
||||
|
||||
vmulps(xmm1, xmm0, xmm3);
|
||||
vmovdqa(ptr[&m_env.d4.z], xmm1);
|
||||
|
||||
for(int i = 0; i < 4; i++)
|
||||
{
|
||||
// m_env.d[i].z = dz * m_shift[i];
|
||||
|
||||
vmulps(xmm1, xmm0, Xmm(4 + i));
|
||||
vmovdqa(ptr[&m_env.d[i].z], xmm1);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if(m_en.z)
|
||||
else
|
||||
{
|
||||
// GSVector4 dz = p.zzzz();
|
||||
// GSVector4 p = vertices[0].p;
|
||||
|
||||
shufps(xmm0, xmm0, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
vmovaps(xmm0, ptr[ecx + 16]);
|
||||
|
||||
// m_env.d4.z = dz * 4.0f;
|
||||
|
||||
movaps(xmm1, xmm0);
|
||||
mulps(xmm1, xmm3);
|
||||
movdqa(xmmword[&m_env.d4.z], xmm1);
|
||||
|
||||
for(int i = 0; i < 4; i++)
|
||||
if(m_en.f)
|
||||
{
|
||||
// m_env.d[i].z = dz * m_shift[i];
|
||||
// m_env.p.f = GSVector4i(p).zzzzh().zzzz();
|
||||
|
||||
movaps(xmm1, xmm0);
|
||||
mulps(xmm1, Xmm(4 + i));
|
||||
movdqa(xmmword[&m_env.d[i].z], xmm1);
|
||||
vcvttps2dq(xmm1, xmm0);
|
||||
vpshufhw(xmm1, xmm1, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
vpshufd(xmm1, xmm1, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
vmovdqa(ptr[&m_env.p.f], xmm1);
|
||||
}
|
||||
|
||||
if(m_en.z)
|
||||
{
|
||||
// GSVector4 z = p.zzzz();
|
||||
|
||||
vshufps(xmm0, xmm0, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
|
||||
if(m_sel.zoverflow)
|
||||
{
|
||||
// m_env.p.z = (GSVector4i(z * 0.5f) << 1) | (GSVector4i(z) & GSVector4i::x00000001());
|
||||
|
||||
static const float half = 0.5f;
|
||||
|
||||
vmovss(xmm1, dword[&half]);
|
||||
vshufps(xmm1, xmm1, _MM_SHUFFLE(0, 0, 0, 0));
|
||||
vmulps(xmm1, xmm0);
|
||||
vcvttps2dq(xmm1, xmm1);
|
||||
vpslld(xmm1, 1);
|
||||
|
||||
vcvttps2dq(xmm0, xmm0);
|
||||
vpcmpeqd(xmm2, xmm2);
|
||||
vpsrld(xmm2, 31);
|
||||
vpand(xmm0, xmm2);
|
||||
|
||||
vpor(xmm0, xmm1);
|
||||
}
|
||||
else
|
||||
{
|
||||
// m_env.p.z = GSVector4i(z);
|
||||
|
||||
vcvttps2dq(xmm0, xmm0);
|
||||
}
|
||||
|
||||
vmovdqa(ptr[&m_env.p.z], xmm0);
|
||||
}
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
// GSVector4 p = vertices[0].p;
|
||||
|
||||
movaps(xmm0, xmmword[ecx + 16]);
|
||||
|
||||
if(m_en.f)
|
||||
if(!m_sel.sprite)
|
||||
{
|
||||
// m_env.p.f = GSVector4i(p).zzzzh().zzzz();
|
||||
// GSVector4 t = dscan.p;
|
||||
|
||||
movaps(xmm1, xmm0);
|
||||
cvttps2dq(xmm1, xmm1);
|
||||
pshufhw(xmm1, xmm1, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
pshufd(xmm1, xmm1, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
movdqa(xmmword[&m_env.p.f], xmm1);
|
||||
movaps(xmm0, ptr[edx + 16]);
|
||||
|
||||
if(m_en.f)
|
||||
{
|
||||
// GSVector4 df = p.wwww();
|
||||
|
||||
movaps(xmm1, xmm0);
|
||||
shufps(xmm1, xmm1, _MM_SHUFFLE(3, 3, 3, 3));
|
||||
|
||||
// m_env.d4.f = GSVector4i(df * 4.0f).xxzzlh();
|
||||
|
||||
movaps(xmm2, xmm1);
|
||||
mulps(xmm2, xmm3);
|
||||
cvttps2dq(xmm2, xmm2);
|
||||
pshuflw(xmm2, xmm2, _MM_SHUFFLE(2, 2, 0, 0));
|
||||
pshufhw(xmm2, xmm2, _MM_SHUFFLE(2, 2, 0, 0));
|
||||
movdqa(ptr[&m_env.d4.f], xmm2);
|
||||
|
||||
for(int i = 0; i < 4; i++)
|
||||
{
|
||||
// m_env.d[i].f = GSVector4i(df * m_shift[i]).xxzzlh();
|
||||
|
||||
movaps(xmm2, xmm1);
|
||||
mulps(xmm2, Xmm(4 + i));
|
||||
cvttps2dq(xmm2, xmm2);
|
||||
pshuflw(xmm2, xmm2, _MM_SHUFFLE(2, 2, 0, 0));
|
||||
pshufhw(xmm2, xmm2, _MM_SHUFFLE(2, 2, 0, 0));
|
||||
movdqa(ptr[&m_env.d[i].f], xmm2);
|
||||
}
|
||||
}
|
||||
|
||||
if(m_en.z)
|
||||
{
|
||||
// GSVector4 dz = p.zzzz();
|
||||
|
||||
shufps(xmm0, xmm0, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
|
||||
// m_env.d4.z = dz * 4.0f;
|
||||
|
||||
movaps(xmm1, xmm0);
|
||||
mulps(xmm1, xmm3);
|
||||
movdqa(ptr[&m_env.d4.z], xmm1);
|
||||
|
||||
for(int i = 0; i < 4; i++)
|
||||
{
|
||||
// m_env.d[i].z = dz * m_shift[i];
|
||||
|
||||
movaps(xmm1, xmm0);
|
||||
mulps(xmm1, Xmm(4 + i));
|
||||
movdqa(ptr[&m_env.d[i].z], xmm1);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if(m_en.z)
|
||||
else
|
||||
{
|
||||
// GSVector4 z = p.zzzz();
|
||||
// GSVector4 p = vertices[0].p;
|
||||
|
||||
shufps(xmm0, xmm0, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
movaps(xmm0, ptr[ecx + 16]);
|
||||
|
||||
if(m_sel.zoverflow)
|
||||
if(m_en.f)
|
||||
{
|
||||
// m_env.p.z = (GSVector4i(z * 0.5f) << 1) | (GSVector4i(z) & GSVector4i::x00000001());
|
||||
// m_env.p.f = GSVector4i(p).zzzzh().zzzz();
|
||||
|
||||
static const float half = 0.5f;
|
||||
|
||||
movss(xmm1, dword[&half]);
|
||||
shufps(xmm1, xmm1, _MM_SHUFFLE(0, 0, 0, 0));
|
||||
mulps(xmm1, xmm0);
|
||||
cvttps2dq(xmm1, xmm1);
|
||||
pslld(xmm1, 1);
|
||||
|
||||
cvttps2dq(xmm0, xmm0);
|
||||
pcmpeqd(xmm2, xmm2);
|
||||
psrld(xmm2, 31);
|
||||
pand(xmm0, xmm2);
|
||||
|
||||
por(xmm0, xmm1);
|
||||
}
|
||||
else
|
||||
{
|
||||
// m_env.p.z = GSVector4i(z);
|
||||
|
||||
cvttps2dq(xmm0, xmm0);
|
||||
cvttps2dq(xmm1, xmm0);
|
||||
pshufhw(xmm1, xmm1, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
pshufd(xmm1, xmm1, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
movdqa(ptr[&m_env.p.f], xmm1);
|
||||
}
|
||||
|
||||
movdqa(xmmword[&m_env.p.z], xmm0);
|
||||
if(m_en.z)
|
||||
{
|
||||
// GSVector4 z = p.zzzz();
|
||||
|
||||
shufps(xmm0, xmm0, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
|
||||
if(m_sel.zoverflow)
|
||||
{
|
||||
// m_env.p.z = (GSVector4i(z * 0.5f) << 1) | (GSVector4i(z) & GSVector4i::x00000001());
|
||||
|
||||
static const float half = 0.5f;
|
||||
|
||||
movss(xmm1, dword[&half]);
|
||||
shufps(xmm1, xmm1, _MM_SHUFFLE(0, 0, 0, 0));
|
||||
mulps(xmm1, xmm0);
|
||||
cvttps2dq(xmm1, xmm1);
|
||||
pslld(xmm1, 1);
|
||||
|
||||
cvttps2dq(xmm0, xmm0);
|
||||
pcmpeqd(xmm2, xmm2);
|
||||
psrld(xmm2, 31);
|
||||
pand(xmm0, xmm2);
|
||||
|
||||
por(xmm0, xmm1);
|
||||
}
|
||||
else
|
||||
{
|
||||
// m_env.p.z = GSVector4i(z);
|
||||
|
||||
cvttps2dq(xmm0, xmm0);
|
||||
}
|
||||
|
||||
movdqa(ptr[&m_env.p.z], xmm0);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -186,64 +301,129 @@ void GSSetupPrimCodeGenerator::Texture()
|
||||
return;
|
||||
}
|
||||
|
||||
// GSVector4 t = dscan.t;
|
||||
|
||||
movaps(xmm0, xmmword[edx + 32]);
|
||||
|
||||
movaps(xmm1, xmm0);
|
||||
mulps(xmm1, xmm3);
|
||||
|
||||
if(m_sel.fst)
|
||||
if(m_cpu.has(util::Cpu::tAVX))
|
||||
{
|
||||
// m_env.d4.st = GSVector4i(t * 4.0f);
|
||||
// GSVector4 t = dscan.t;
|
||||
|
||||
cvttps2dq(xmm1, xmm1);
|
||||
movdqa(xmmword[&m_env.d4.st], xmm1);
|
||||
vmovaps(xmm0, ptr[edx + 32]);
|
||||
|
||||
vmulps(xmm1, xmm0, xmm3);
|
||||
|
||||
if(m_sel.fst)
|
||||
{
|
||||
// m_env.d4.st = GSVector4i(t * 4.0f);
|
||||
|
||||
vcvttps2dq(xmm1, xmm1);
|
||||
vmovdqa(ptr[&m_env.d4.st], xmm1);
|
||||
}
|
||||
else
|
||||
{
|
||||
// m_env.d4.stq = t * 4.0f;
|
||||
|
||||
vmovaps(ptr[&m_env.d4.stq], xmm1);
|
||||
}
|
||||
|
||||
for(int j = 0, k = m_sel.fst ? 2 : 3; j < k; j++)
|
||||
{
|
||||
// GSVector4 ds = t.xxxx();
|
||||
// GSVector4 dt = t.yyyy();
|
||||
// GSVector4 dq = t.zzzz();
|
||||
|
||||
vshufps(xmm1, xmm0, xmm0, (uint8)_MM_SHUFFLE(j, j, j, j));
|
||||
|
||||
for(int i = 0; i < 4; i++)
|
||||
{
|
||||
// GSVector4 v = ds/dt * m_shift[i];
|
||||
|
||||
vmulps(xmm2, xmm1, Xmm(4 + i));
|
||||
|
||||
if(m_sel.fst)
|
||||
{
|
||||
// m_env.d[i].si/ti = GSVector4i(v);
|
||||
|
||||
vcvttps2dq(xmm2, xmm2);
|
||||
|
||||
switch(j)
|
||||
{
|
||||
case 0: vmovdqa(ptr[&m_env.d[i].si], xmm2); break;
|
||||
case 1: vmovdqa(ptr[&m_env.d[i].ti], xmm2); break;
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
// m_env.d[i].s/t/q = v;
|
||||
|
||||
switch(j)
|
||||
{
|
||||
case 0: vmovaps(ptr[&m_env.d[i].s], xmm2); break;
|
||||
case 1: vmovaps(ptr[&m_env.d[i].t], xmm2); break;
|
||||
case 2: vmovaps(ptr[&m_env.d[i].q], xmm2); break;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
// m_env.d4.stq = t * 4.0f;
|
||||
// GSVector4 t = dscan.t;
|
||||
|
||||
movaps(xmmword[&m_env.d4.stq], xmm1);
|
||||
}
|
||||
|
||||
for(int j = 0, k = m_sel.fst ? 2 : 3; j < k; j++)
|
||||
{
|
||||
// GSVector4 ds = t.xxxx();
|
||||
// GSVector4 dt = t.yyyy();
|
||||
// GSVector4 dq = t.zzzz();
|
||||
movaps(xmm0, ptr[edx + 32]);
|
||||
|
||||
movaps(xmm1, xmm0);
|
||||
shufps(xmm1, xmm1, (uint8)_MM_SHUFFLE(j, j, j, j));
|
||||
mulps(xmm1, xmm3);
|
||||
|
||||
for(int i = 0; i < 4; i++)
|
||||
if(m_sel.fst)
|
||||
{
|
||||
// GSVector4 v = ds/dt * m_shift[i];
|
||||
// m_env.d4.st = GSVector4i(t * 4.0f);
|
||||
|
||||
movaps(xmm2, xmm1);
|
||||
mulps(xmm2, Xmm(4 + i));
|
||||
cvttps2dq(xmm1, xmm1);
|
||||
movdqa(ptr[&m_env.d4.st], xmm1);
|
||||
}
|
||||
else
|
||||
{
|
||||
// m_env.d4.stq = t * 4.0f;
|
||||
|
||||
if(m_sel.fst)
|
||||
movaps(ptr[&m_env.d4.stq], xmm1);
|
||||
}
|
||||
|
||||
for(int j = 0, k = m_sel.fst ? 2 : 3; j < k; j++)
|
||||
{
|
||||
// GSVector4 ds = t.xxxx();
|
||||
// GSVector4 dt = t.yyyy();
|
||||
// GSVector4 dq = t.zzzz();
|
||||
|
||||
movaps(xmm1, xmm0);
|
||||
shufps(xmm1, xmm1, (uint8)_MM_SHUFFLE(j, j, j, j));
|
||||
|
||||
for(int i = 0; i < 4; i++)
|
||||
{
|
||||
// m_env.d[i].si/ti = GSVector4i(v);
|
||||
// GSVector4 v = ds/dt * m_shift[i];
|
||||
|
||||
cvttps2dq(xmm2, xmm2);
|
||||
movaps(xmm2, xmm1);
|
||||
mulps(xmm2, Xmm(4 + i));
|
||||
|
||||
switch(j)
|
||||
if(m_sel.fst)
|
||||
{
|
||||
case 0: movdqa(xmmword[&m_env.d[i].si], xmm2); break;
|
||||
case 1: movdqa(xmmword[&m_env.d[i].ti], xmm2); break;
|
||||
// m_env.d[i].si/ti = GSVector4i(v);
|
||||
|
||||
cvttps2dq(xmm2, xmm2);
|
||||
|
||||
switch(j)
|
||||
{
|
||||
case 0: movdqa(ptr[&m_env.d[i].si], xmm2); break;
|
||||
case 1: movdqa(ptr[&m_env.d[i].ti], xmm2); break;
|
||||
}
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
// m_env.d[i].s/t/q = v;
|
||||
|
||||
switch(j)
|
||||
else
|
||||
{
|
||||
case 0: movaps(xmmword[&m_env.d[i].s], xmm2); break;
|
||||
case 1: movaps(xmmword[&m_env.d[i].t], xmm2); break;
|
||||
case 2: movaps(xmmword[&m_env.d[i].q], xmm2); break;
|
||||
// m_env.d[i].s/t/q = v;
|
||||
|
||||
switch(j)
|
||||
{
|
||||
case 0: movaps(ptr[&m_env.d[i].s], xmm2); break;
|
||||
case 1: movaps(ptr[&m_env.d[i].t], xmm2); break;
|
||||
case 2: movaps(ptr[&m_env.d[i].q], xmm2); break;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -257,113 +437,217 @@ void GSSetupPrimCodeGenerator::Color()
|
||||
return;
|
||||
}
|
||||
|
||||
if(m_sel.iip)
|
||||
if(m_cpu.has(util::Cpu::tAVX))
|
||||
{
|
||||
// GSVector4 c = dscan.c;
|
||||
|
||||
movaps(xmm0, xmmword[edx]);
|
||||
movaps(xmm1, xmm0);
|
||||
|
||||
// m_env.d4.c = GSVector4i(c * 4.0f).xzyw().ps32();
|
||||
|
||||
movaps(xmm2, xmm0);
|
||||
mulps(xmm2, xmm3);
|
||||
cvttps2dq(xmm2, xmm2);
|
||||
pshufd(xmm2, xmm2, _MM_SHUFFLE(3, 1, 2, 0));
|
||||
packssdw(xmm2, xmm2);
|
||||
movdqa(xmmword[&m_env.d4.c], xmm2);
|
||||
|
||||
// xmm3 is not needed anymore
|
||||
|
||||
// GSVector4 dr = c.xxxx();
|
||||
// GSVector4 db = c.zzzz();
|
||||
|
||||
shufps(xmm0, xmm0, _MM_SHUFFLE(0, 0, 0, 0));
|
||||
shufps(xmm1, xmm1, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
|
||||
for(int i = 0; i < 4; i++)
|
||||
if(m_sel.iip)
|
||||
{
|
||||
// GSVector4i r = GSVector4i(dr * m_shift[i]).ps32();
|
||||
// GSVector4 c = dscan.c;
|
||||
|
||||
movaps(xmm2, xmm0);
|
||||
mulps(xmm2, Xmm(4 + i));
|
||||
cvttps2dq(xmm2, xmm2);
|
||||
packssdw(xmm2, xmm2);
|
||||
vmovaps(xmm0, ptr[edx]);
|
||||
|
||||
// GSVector4i b = GSVector4i(db * m_shift[i]).ps32();
|
||||
// m_env.d4.c = GSVector4i(c * 4.0f).xzyw().ps32();
|
||||
|
||||
movaps(xmm3, xmm1);
|
||||
mulps(xmm3, Xmm(4 + i));
|
||||
cvttps2dq(xmm3, xmm3);
|
||||
packssdw(xmm3, xmm3);
|
||||
vmulps(xmm1, xmm0, xmm3);
|
||||
vcvttps2dq(xmm1, xmm1);
|
||||
vpshufd(xmm1, xmm1, _MM_SHUFFLE(3, 1, 2, 0));
|
||||
vpackssdw(xmm1, xmm1);
|
||||
vmovdqa(ptr[&m_env.d4.c], xmm1);
|
||||
|
||||
// m_env.d[i].rb = r.upl16(b);
|
||||
// xmm3 is not needed anymore
|
||||
|
||||
punpcklwd(xmm2, xmm3);
|
||||
movdqa(xmmword[&m_env.d[i].rb], xmm2);
|
||||
// GSVector4 dr = c.xxxx();
|
||||
// GSVector4 db = c.zzzz();
|
||||
|
||||
vshufps(xmm2, xmm0, xmm0, _MM_SHUFFLE(0, 0, 0, 0));
|
||||
vshufps(xmm3, xmm0, xmm0, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
|
||||
for(int i = 0; i < 4; i++)
|
||||
{
|
||||
// GSVector4i r = GSVector4i(dr * m_shift[i]).ps32();
|
||||
|
||||
vmulps(xmm0, xmm2, Xmm(4 + i));
|
||||
vcvttps2dq(xmm0, xmm0);
|
||||
vpackssdw(xmm0, xmm0);
|
||||
|
||||
// GSVector4i b = GSVector4i(db * m_shift[i]).ps32();
|
||||
|
||||
vmulps(xmm1, xmm3, Xmm(4 + i));
|
||||
vcvttps2dq(xmm1, xmm1);
|
||||
vpackssdw(xmm1, xmm1);
|
||||
|
||||
// m_env.d[i].rb = r.upl16(b);
|
||||
|
||||
vpunpcklwd(xmm0, xmm1);
|
||||
vmovdqa(ptr[&m_env.d[i].rb], xmm0);
|
||||
}
|
||||
|
||||
// GSVector4 c = dscan.c;
|
||||
|
||||
vmovaps(xmm0, ptr[edx]); // not enough regs, have to reload it
|
||||
|
||||
// GSVector4 dg = c.yyyy();
|
||||
// GSVector4 da = c.wwww();
|
||||
|
||||
vshufps(xmm2, xmm0, xmm0, _MM_SHUFFLE(1, 1, 1, 1));
|
||||
vshufps(xmm3, xmm0, xmm0, _MM_SHUFFLE(3, 3, 3, 3));
|
||||
|
||||
for(int i = 0; i < 4; i++)
|
||||
{
|
||||
// GSVector4i g = GSVector4i(dg * m_shift[i]).ps32();
|
||||
|
||||
vmulps(xmm0, xmm2, Xmm(4 + i));
|
||||
vcvttps2dq(xmm0, xmm0);
|
||||
vpackssdw(xmm0, xmm0);
|
||||
|
||||
// GSVector4i a = GSVector4i(da * m_shift[i]).ps32();
|
||||
|
||||
vmulps(xmm1, xmm3, Xmm(4 + i));
|
||||
vcvttps2dq(xmm1, xmm1);
|
||||
vpackssdw(xmm1, xmm1);
|
||||
|
||||
// m_env.d[i].ga = g.upl16(a);
|
||||
|
||||
vpunpcklwd(xmm0, xmm1);
|
||||
vmovdqa(ptr[&m_env.d[i].ga], xmm0);
|
||||
}
|
||||
}
|
||||
|
||||
// GSVector4 c = dscan.c;
|
||||
|
||||
movaps(xmm0, xmmword[edx]); // not enough regs, have to reload it
|
||||
movaps(xmm1, xmm0);
|
||||
|
||||
// GSVector4 dg = c.yyyy();
|
||||
// GSVector4 da = c.wwww();
|
||||
|
||||
shufps(xmm0, xmm0, _MM_SHUFFLE(1, 1, 1, 1));
|
||||
shufps(xmm1, xmm1, _MM_SHUFFLE(3, 3, 3, 3));
|
||||
|
||||
for(int i = 0; i < 4; i++)
|
||||
else
|
||||
{
|
||||
// GSVector4i g = GSVector4i(dg * m_shift[i]).ps32();
|
||||
// GSVector4i c = GSVector4i(vertices[0].c);
|
||||
|
||||
movaps(xmm2, xmm0);
|
||||
mulps(xmm2, Xmm(4 + i));
|
||||
cvttps2dq(xmm2, xmm2);
|
||||
packssdw(xmm2, xmm2);
|
||||
vcvttps2dq(xmm0, ptr[ecx]);
|
||||
|
||||
// GSVector4i a = GSVector4i(da * m_shift[i]).ps32();
|
||||
// c = c.upl16(c.zwxy());
|
||||
|
||||
movaps(xmm3, xmm1);
|
||||
mulps(xmm3, Xmm(4 + i));
|
||||
cvttps2dq(xmm3, xmm3);
|
||||
packssdw(xmm3, xmm3);
|
||||
vpshufd(xmm1, xmm0, _MM_SHUFFLE(1, 0, 3, 2));
|
||||
vpunpcklwd(xmm0, xmm1);
|
||||
|
||||
// m_env.d[i].ga = g.upl16(a);
|
||||
// if(!tme) c = c.srl16(7);
|
||||
|
||||
punpcklwd(xmm2, xmm3);
|
||||
movdqa(xmmword[&m_env.d[i].ga], xmm2);
|
||||
if(m_sel.tfx == TFX_NONE)
|
||||
{
|
||||
vpsrlw(xmm0, 7);
|
||||
}
|
||||
|
||||
// m_env.c.rb = c.xxxx();
|
||||
// m_env.c.ga = c.zzzz();
|
||||
|
||||
vpshufd(xmm1, xmm0, _MM_SHUFFLE(0, 0, 0, 0));
|
||||
vpshufd(xmm2, xmm0, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
|
||||
vmovdqa(ptr[&m_env.c.rb], xmm1);
|
||||
vmovdqa(ptr[&m_env.c.ga], xmm2);
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
// GSVector4i c = GSVector4i(vertices[0].c);
|
||||
|
||||
movaps(xmm0, xmmword[ecx]);
|
||||
cvttps2dq(xmm0, xmm0);
|
||||
|
||||
// c = c.upl16(c.zwxy());
|
||||
|
||||
movdqa(xmm1, xmm0);
|
||||
pshufd(xmm1, xmm1, _MM_SHUFFLE(1, 0, 3, 2));
|
||||
punpcklwd(xmm0, xmm1);
|
||||
|
||||
// if(!tme) c = c.srl16(7);
|
||||
|
||||
if(m_sel.tfx == TFX_NONE)
|
||||
if(m_sel.iip)
|
||||
{
|
||||
psrlw(xmm0, 7);
|
||||
// GSVector4 c = dscan.c;
|
||||
|
||||
movaps(xmm0, ptr[edx]);
|
||||
movaps(xmm1, xmm0);
|
||||
|
||||
// m_env.d4.c = GSVector4i(c * 4.0f).xzyw().ps32();
|
||||
|
||||
movaps(xmm2, xmm0);
|
||||
mulps(xmm2, xmm3);
|
||||
cvttps2dq(xmm2, xmm2);
|
||||
pshufd(xmm2, xmm2, _MM_SHUFFLE(3, 1, 2, 0));
|
||||
packssdw(xmm2, xmm2);
|
||||
movdqa(ptr[&m_env.d4.c], xmm2);
|
||||
|
||||
// xmm3 is not needed anymore
|
||||
|
||||
// GSVector4 dr = c.xxxx();
|
||||
// GSVector4 db = c.zzzz();
|
||||
|
||||
shufps(xmm0, xmm0, _MM_SHUFFLE(0, 0, 0, 0));
|
||||
shufps(xmm1, xmm1, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
|
||||
for(int i = 0; i < 4; i++)
|
||||
{
|
||||
// GSVector4i r = GSVector4i(dr * m_shift[i]).ps32();
|
||||
|
||||
movaps(xmm2, xmm0);
|
||||
mulps(xmm2, Xmm(4 + i));
|
||||
cvttps2dq(xmm2, xmm2);
|
||||
packssdw(xmm2, xmm2);
|
||||
|
||||
// GSVector4i b = GSVector4i(db * m_shift[i]).ps32();
|
||||
|
||||
movaps(xmm3, xmm1);
|
||||
mulps(xmm3, Xmm(4 + i));
|
||||
cvttps2dq(xmm3, xmm3);
|
||||
packssdw(xmm3, xmm3);
|
||||
|
||||
// m_env.d[i].rb = r.upl16(b);
|
||||
|
||||
punpcklwd(xmm2, xmm3);
|
||||
movdqa(ptr[&m_env.d[i].rb], xmm2);
|
||||
}
|
||||
|
||||
// GSVector4 c = dscan.c;
|
||||
|
||||
movaps(xmm0, ptr[edx]); // not enough regs, have to reload it
|
||||
movaps(xmm1, xmm0);
|
||||
|
||||
// GSVector4 dg = c.yyyy();
|
||||
// GSVector4 da = c.wwww();
|
||||
|
||||
shufps(xmm0, xmm0, _MM_SHUFFLE(1, 1, 1, 1));
|
||||
shufps(xmm1, xmm1, _MM_SHUFFLE(3, 3, 3, 3));
|
||||
|
||||
for(int i = 0; i < 4; i++)
|
||||
{
|
||||
// GSVector4i g = GSVector4i(dg * m_shift[i]).ps32();
|
||||
|
||||
movaps(xmm2, xmm0);
|
||||
mulps(xmm2, Xmm(4 + i));
|
||||
cvttps2dq(xmm2, xmm2);
|
||||
packssdw(xmm2, xmm2);
|
||||
|
||||
// GSVector4i a = GSVector4i(da * m_shift[i]).ps32();
|
||||
|
||||
movaps(xmm3, xmm1);
|
||||
mulps(xmm3, Xmm(4 + i));
|
||||
cvttps2dq(xmm3, xmm3);
|
||||
packssdw(xmm3, xmm3);
|
||||
|
||||
// m_env.d[i].ga = g.upl16(a);
|
||||
|
||||
punpcklwd(xmm2, xmm3);
|
||||
movdqa(ptr[&m_env.d[i].ga], xmm2);
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
// GSVector4i c = GSVector4i(vertices[0].c);
|
||||
|
||||
// m_env.c.rb = c.xxxx();
|
||||
// m_env.c.ga = c.zzzz();
|
||||
movaps(xmm0, ptr[ecx]);
|
||||
cvttps2dq(xmm0, xmm0);
|
||||
|
||||
movdqa(xmm1, xmm0);
|
||||
pshufd(xmm0, xmm0, _MM_SHUFFLE(0, 0, 0, 0));
|
||||
pshufd(xmm1, xmm1, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
movdqa(xmmword[&m_env.c.rb], xmm0);
|
||||
movdqa(xmmword[&m_env.c.ga], xmm1);
|
||||
// c = c.upl16(c.zwxy());
|
||||
|
||||
pshufd(xmm1, xmm0, _MM_SHUFFLE(1, 0, 3, 2));
|
||||
punpcklwd(xmm0, xmm1);
|
||||
|
||||
// if(!tme) c = c.srl16(7);
|
||||
|
||||
if(m_sel.tfx == TFX_NONE)
|
||||
{
|
||||
psrlw(xmm0, 7);
|
||||
}
|
||||
|
||||
// m_env.c.rb = c.xxxx();
|
||||
// m_env.c.ga = c.zzzz();
|
||||
|
||||
pshufd(xmm1, xmm0, _MM_SHUFFLE(0, 0, 0, 0));
|
||||
pshufd(xmm2, xmm0, _MM_SHUFFLE(2, 2, 2, 2));
|
||||
|
||||
movdqa(ptr[&m_env.c.rb], xmm1);
|
||||
movdqa(ptr[&m_env.c.ga], xmm2);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user