void GPUDrawScanlineCodeGenerator::blend8(const Xmm& a, const Xmm& b)
{
if(m_cpu.has(util::Cpu::tSSE41))
{
pblendvb(a, b);
}
else
{
blend(a, b, xmm0);
}
}
void GSDrawScanlineCodeGenerator::blend8r(const Xmm& b, const Xmm& a)
{
if(m_cpu.has(util::Cpu::tSSE41))
{
pblendvb(a, b);
movdqa(b, a);
}
else
{
blendr(b, a, xmm0);
}
}
void GSDrawScanlineCodeGenerator::blend8(const Xmm& a, const Xmm& b)
{
#if _M_SSE >= 0x500
vpblendvb(a, a, b, xmm0);
#elif _M_SSE >= 0x401
pblendvb(a, b);
#else
blend(a, b, xmm0);
#endif
}
void GSDrawScanlineCodeGenerator::blend8r(const Xmm& b, const Xmm& a)
{
#if _M_SSE >= 0x500
vpblendvb(b, a, b, xmm0);
#elif _M_SSE >= 0x401
pblendvb(a, b);
movdqa(b, a);
#else
blendr(b, a, xmm0);
#endif
}
void GSDrawScanlineCodeGenerator::Wrap(const Xmm& uv0, const Xmm& uv1)
{
// xmm0, xmm1, xmm4, xmm5, xmm6 = free
int wms_clamp = ((m_sel.wms + 1) >> 1) & 1;
int wmt_clamp = ((m_sel.wmt + 1) >> 1) & 1;
int region = ((m_sel.wms | m_sel.wmt) >> 1) & 1;
if(wms_clamp == wmt_clamp)
{
if(wms_clamp)
{
if(region)
{
movdqa(xmm4, xmmword[&m_env.t.min]);
pmaxsw(uv0, xmm4);
pmaxsw(uv1, xmm4);
}
else
{
pxor(xmm0, xmm0);
pmaxsw(uv0, xmm0);
pmaxsw(uv1, xmm0);
}
movdqa(xmm5, xmmword[&m_env.t.max]);
pminsw(uv0, xmm5);
pminsw(uv1, xmm5);
}
else
{
movdqa(xmm4, xmmword[&m_env.t.min]);
pand(uv0, xmm4);
pand(uv1, xmm4);
if(region)
{
movdqa(xmm5, xmmword[&m_env.t.max]);
por(uv0, xmm5);
por(uv1, xmm5);
}
}
}
else
{
movdqa(xmm1, uv0);
movdqa(xmm6, uv1);
movdqa(xmm4, xmmword[&m_env.t.min]);
movdqa(xmm5, xmmword[&m_env.t.max]);
// GSVector4i clamp = t.sat_i16(m_env.t.min, m_env.t.max);
pmaxsw(uv0, xmm4);
pmaxsw(uv1, xmm4);
pminsw(uv0, xmm5);
pminsw(uv1, xmm5);
// GSVector4i repeat = (t & m_env.t.min) | m_env.t.max;
pand(xmm1, xmm4);
pand(xmm6, xmm4);
if(region)
{
por(xmm1, xmm5);
por(xmm6, xmm5);
}
// clamp.blend8(repeat, m_env.t.mask);
if(m_cpu.has(util::Cpu::tSSE41))
{
movdqa(xmm0, xmmword[&m_env.t.mask]);
pblendvb(uv0, xmm1);
pblendvb(uv1, xmm6);
}
else
{
movdqa(xmm0, xmmword[&m_env.t.invmask]);
movdqa(xmm4, xmm0);
pand(uv0, xmm0);
pandn(xmm0, xmm1);
por(uv0, xmm0);
pand(uv1, xmm4);
pandn(xmm4, xmm6);
por(uv1, xmm4);
}
}
}
