2010-03-19 00:31:15 +00:00
|
|
|
/* ZeroGS KOSMOS
|
|
|
|
* Copyright (C) 2005-2006 Gabest/zerofrog@gmail.com
|
|
|
|
* http://www.gabest.org
|
|
|
|
*
|
|
|
|
* This Program is free software; you can redistribute it and/or modify
|
|
|
|
* it under the terms of the GNU General Public License as published by
|
|
|
|
* the Free Software Foundation; either version 2, or (at your option)
|
|
|
|
* any later version.
|
2010-04-25 00:31:27 +00:00
|
|
|
*
|
2010-03-19 00:31:15 +00:00
|
|
|
* This Program is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
* GNU General Public License for more details.
|
2010-04-25 00:31:27 +00:00
|
|
|
*
|
2010-03-19 00:31:15 +00:00
|
|
|
* You should have received a copy of the GNU General Public License
|
|
|
|
* along with GNU Make; see the file COPYING. If not, write to
|
2010-04-25 00:31:27 +00:00
|
|
|
* the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.
|
2010-03-19 00:31:15 +00:00
|
|
|
* http://www.gnu.org/copyleft/gpl.html
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "GS.h"
|
|
|
|
#include "Mem.h"
|
|
|
|
#include "x86.h"
|
|
|
|
|
2010-03-31 12:27:36 +00:00
|
|
|
#if defined(ZEROGS_SSE2)
|
2010-03-19 00:31:15 +00:00
|
|
|
#include <xmmintrin.h>
|
|
|
|
#include <emmintrin.h>
|
|
|
|
#endif
|
|
|
|
|
|
|
|
// swizzling
|
|
|
|
#define _FrameSwizzleBlock(type, transfer, transfer16, incsrc) \
|
|
|
|
/* FrameSwizzleBlock32 */ \
|
|
|
|
void __fastcall FrameSwizzleBlock32##type##c(u32* dst, u32* src, int srcpitch, u32 WriteMask) \
|
|
|
|
{ \
|
|
|
|
u32* d = &g_columnTable32[0][0]; \
|
|
|
|
\
|
|
|
|
if( WriteMask == 0xffffffff ) { \
|
|
|
|
for(int i = 0; i < 8; ++i, d += 8) { \
|
|
|
|
for(int j = 0; j < 8; ++j) { \
|
|
|
|
dst[d[j]] = (transfer); \
|
|
|
|
} \
|
|
|
|
src += srcpitch << incsrc; \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
else { \
|
|
|
|
for(int i = 0; i < 8; ++i, d += 8) { \
|
|
|
|
for(int j = 0; j < 8; ++j) { \
|
|
|
|
dst[d[j]] = ((transfer)&WriteMask)|(dst[d[j]]&~WriteMask); \
|
|
|
|
} \
|
|
|
|
src += srcpitch << incsrc; \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
\
|
|
|
|
void __fastcall FrameSwizzleBlock24##type##c(u32* dst, u32* src, int srcpitch, u32 WriteMask) \
|
|
|
|
{ \
|
|
|
|
u32* d = &g_columnTable32[0][0]; \
|
|
|
|
\
|
|
|
|
if( WriteMask == 0xffffffff ) { \
|
|
|
|
for(int i = 0; i < 8; ++i, d += 8) { \
|
|
|
|
for(int j = 0; j < 8; ++j) { \
|
|
|
|
dst[d[j]] = (transfer); \
|
|
|
|
} \
|
|
|
|
src += srcpitch << incsrc; \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
else { \
|
|
|
|
for(int i = 0; i < 8; ++i, d += 8) { \
|
|
|
|
for(int j = 0; j < 8; ++j) { \
|
|
|
|
dst[d[j]] = ((transfer)&WriteMask)|(dst[d[j]]&~WriteMask); \
|
|
|
|
} \
|
|
|
|
src += srcpitch << incsrc; \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
\
|
|
|
|
/* FrameSwizzleBlock16 */ \
|
|
|
|
void __fastcall FrameSwizzleBlock16##type##c(u16* dst, u32* src, int srcpitch, u32 WriteMask) \
|
|
|
|
{ \
|
|
|
|
u32* d = &g_columnTable16[0][0]; \
|
|
|
|
\
|
|
|
|
if( WriteMask == 0xffff ) { \
|
|
|
|
for(int i = 0; i < 8; ++i, d += 16) { \
|
|
|
|
for(int j = 0; j < 16; ++j) { \
|
|
|
|
u32 temp = (transfer); \
|
|
|
|
dst[d[j]] = RGBA32to16(temp); \
|
|
|
|
} \
|
|
|
|
src += srcpitch << incsrc; \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
else { \
|
|
|
|
for(int i = 0; i < 8; ++i, d += 16) { \
|
|
|
|
for(int j = 0; j < 16; ++j) { \
|
|
|
|
u32 temp = (transfer); \
|
|
|
|
u32 dsrc = RGBA32to16(temp); \
|
|
|
|
dst[d[j]] = (dsrc&WriteMask)|(dst[d[j]]&~WriteMask); \
|
|
|
|
} \
|
|
|
|
src += srcpitch << incsrc; \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
\
|
|
|
|
/* Frame16SwizzleBlock32 */ \
|
|
|
|
void __fastcall Frame16SwizzleBlock32##type##c(u32* dst, Vector_16F* src, int srcpitch, u32 WriteMask) \
|
|
|
|
{ \
|
|
|
|
u32* d = &g_columnTable32[0][0]; \
|
|
|
|
\
|
|
|
|
if( WriteMask == 0xffffffff ) { \
|
|
|
|
for(int i = 0; i < 8; ++i, d += 8) { \
|
|
|
|
for(int j = 0; j < 8; ++j) { \
|
|
|
|
Vector_16F dsrc16 = (transfer16); \
|
|
|
|
dst[d[j]] = Float16ToARGB(dsrc16); \
|
|
|
|
} \
|
|
|
|
src += srcpitch << incsrc; \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
else { \
|
|
|
|
for(int i = 0; i < 8; ++i, d += 8) { \
|
|
|
|
for(int j = 0; j < 8; ++j) { \
|
|
|
|
Vector_16F dsrc16 = (transfer16); \
|
|
|
|
u32 dsrc = Float16ToARGB(dsrc16); \
|
|
|
|
dst[d[j]] = (dsrc&WriteMask)|(dst[d[j]]&~WriteMask); \
|
|
|
|
} \
|
|
|
|
src += srcpitch << incsrc; \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
\
|
|
|
|
/* Frame16SwizzleBlock32Z */ \
|
|
|
|
void __fastcall Frame16SwizzleBlock32Z##type##c(u32* dst, Vector_16F* src, int srcpitch, u32 WriteMask) \
|
|
|
|
{ \
|
|
|
|
u32* d = &g_columnTable32[0][0]; \
|
|
|
|
if( WriteMask == 0xffffffff ) { /* breaks KH text if not checked */ \
|
|
|
|
for(int i = 0; i < 8; ++i, d += 8) { \
|
|
|
|
for(int j = 0; j < 8; ++j) { \
|
|
|
|
Vector_16F dsrc16 = (transfer16); \
|
|
|
|
dst[d[j]] = Float16ToARGB_Z(dsrc16); \
|
|
|
|
} \
|
|
|
|
src += srcpitch << incsrc; \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
else { \
|
|
|
|
for(int i = 0; i < 8; ++i, d += 8) { \
|
|
|
|
for(int j = 0; j < 8; ++j) { \
|
|
|
|
Vector_16F dsrc16 = (transfer16); \
|
|
|
|
u32 dsrc = Float16ToARGB_Z(dsrc16); \
|
|
|
|
dst[d[j]] = (dsrc&WriteMask)|(dst[d[j]]&~WriteMask); \
|
|
|
|
} \
|
|
|
|
src += srcpitch << incsrc; \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
\
|
|
|
|
/* Frame16SwizzleBlock16 */ \
|
|
|
|
void __fastcall Frame16SwizzleBlock16##type##c(u16* dst, Vector_16F* src, int srcpitch, u32 WriteMask) \
|
|
|
|
{ \
|
|
|
|
u32* d = &g_columnTable16[0][0]; \
|
|
|
|
\
|
|
|
|
if( (WriteMask&0xfff8f8f8) == 0xfff8f8f8) { \
|
|
|
|
for(int i = 0; i < 8; ++i, d += 16) { \
|
|
|
|
for(int j = 0; j < 16; ++j) { \
|
|
|
|
Vector_16F dsrc16 = (transfer16); \
|
|
|
|
dst[d[j]] = Float16ToARGB16(dsrc16); \
|
|
|
|
} \
|
|
|
|
src += srcpitch << incsrc; \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
else { \
|
|
|
|
for(int i = 0; i < 8; ++i, d += 16) { \
|
|
|
|
for(int j = 0; j < 16; ++j) { \
|
|
|
|
Vector_16F dsrc16 = (transfer16); \
|
|
|
|
u32 dsrc = Float16ToARGB16(dsrc16); \
|
|
|
|
dst[d[j]] = (dsrc&WriteMask)|(dst[d[j]]&~WriteMask); \
|
|
|
|
} \
|
|
|
|
src += srcpitch << incsrc; \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
\
|
|
|
|
/* Frame16SwizzleBlock16Z */ \
|
|
|
|
void __fastcall Frame16SwizzleBlock16Z##type##c(u16* dst, Vector_16F* src, int srcpitch, u32 WriteMask) \
|
|
|
|
{ \
|
|
|
|
u32* d = &g_columnTable16[0][0]; \
|
|
|
|
\
|
|
|
|
for(int i = 0; i < 8; ++i, d += 16) { \
|
|
|
|
for(int j = 0; j < 16; ++j) { \
|
|
|
|
Vector_16F dsrc16 = (transfer16); \
|
|
|
|
dst[d[j]] = Float16ToARGB16_Z(dsrc16); \
|
|
|
|
} \
|
|
|
|
src += srcpitch << incsrc; \
|
|
|
|
} \
|
|
|
|
} \
|
2010-05-01 20:33:53 +00:00
|
|
|
|
2010-03-19 00:31:15 +00:00
|
|
|
_FrameSwizzleBlock(_, src[j], src[j], 0);
|
2010-05-01 20:33:53 +00:00
|
|
|
_FrameSwizzleBlock(A2_, (src[2*j] + src[2*j+1]) >> 1, src[2*j], 0);
|
|
|
|
_FrameSwizzleBlock(A4_, (src[2*j] + src[2*j+1] + src[2*j+srcpitch] + src[2*j+srcpitch+1]) >> 2, src[2*j], 1);
|
2010-03-19 00:31:15 +00:00
|
|
|
|
|
|
|
#ifdef ZEROGS_SSE2
|
|
|
|
|
|
|
|
//void __fastcall WriteCLUT_T16_I8_CSM1_sse2(u32* vm, u32* clut)
|
|
|
|
//{
|
|
|
|
// __asm {
|
|
|
|
// mov eax, vm
|
|
|
|
// mov ecx, clut
|
|
|
|
// mov edx, 8
|
|
|
|
// }
|
|
|
|
//
|
|
|
|
//Extract32x2:
|
|
|
|
// __asm {
|
|
|
|
// movdqa xmm0, qword ptr [eax]
|
|
|
|
// movdqa xmm1, qword ptr [eax+16]
|
|
|
|
// movdqa xmm2, qword ptr [eax+32]
|
|
|
|
// movdqa xmm3, qword ptr [eax+48]
|
|
|
|
//
|
|
|
|
// // rearrange
|
|
|
|
// pshuflw xmm0, xmm0, 0xd8
|
|
|
|
// pshufhw xmm0, xmm0, 0xd8
|
|
|
|
// pshuflw xmm1, xmm1, 0xd8
|
|
|
|
// pshufhw xmm1, xmm1, 0xd8
|
|
|
|
// pshuflw xmm2, xmm2, 0xd8
|
|
|
|
// pshufhw xmm2, xmm2, 0xd8
|
|
|
|
// pshuflw xmm3, xmm3, 0xd8
|
|
|
|
// pshufhw xmm3, xmm3, 0xd8
|
|
|
|
//
|
|
|
|
// movdqa xmm4, xmm0
|
|
|
|
// movdqa xmm6, xmm2
|
|
|
|
//
|
|
|
|
// shufps xmm0, xmm1, 0x88
|
|
|
|
// shufps xmm2, xmm3, 0x88
|
|
|
|
//
|
|
|
|
// shufps xmm4, xmm1, 0xdd
|
|
|
|
// shufps xmm6, xmm3, 0xdd
|
|
|
|
//
|
|
|
|
// pshufd xmm0, xmm0, 0xd8
|
|
|
|
// pshufd xmm2, xmm2, 0xd8
|
|
|
|
// pshufd xmm4, xmm4, 0xd8
|
|
|
|
// pshufd xmm6, xmm6, 0xd8
|
|
|
|
//
|
|
|
|
// // left column
|
|
|
|
// movhlps xmm1, xmm0
|
|
|
|
// movlhps xmm0, xmm2
|
|
|
|
// //movdqa xmm7, [ecx]
|
|
|
|
//
|
|
|
|
// movdqa [ecx], xmm0
|
|
|
|
// shufps xmm1, xmm2, 0xe4
|
|
|
|
// movdqa [ecx+16], xmm1
|
|
|
|
//
|
|
|
|
// // right column
|
|
|
|
// movhlps xmm3, xmm4
|
|
|
|
// movlhps xmm4, xmm6
|
|
|
|
// movdqa [ecx+32], xmm4
|
|
|
|
// shufps xmm3, xmm6, 0xe4
|
|
|
|
// movdqa [ecx+48], xmm3
|
|
|
|
//
|
|
|
|
// add eax, 16*4
|
|
|
|
// add ecx, 16*8
|
|
|
|
// sub edx, 1
|
|
|
|
// cmp edx, 0
|
|
|
|
// jne Extract32x2
|
|
|
|
// }
|
|
|
|
//}
|
|
|
|
|
|
|
|
extern "C" void __fastcall WriteCLUT_T32_I8_CSM1_sse2(u32* vm, u32* clut)
|
|
|
|
{
|
|
|
|
__m128i* src = (__m128i*)vm;
|
|
|
|
__m128i* dst = (__m128i*)clut;
|
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
for (int j = 0; j < 64; j += 32, src += 32, dst += 32)
|
2010-03-19 00:31:15 +00:00
|
|
|
{
|
2010-05-01 20:33:53 +00:00
|
|
|
for (int i = 0; i < 16; i += 4)
|
2010-03-19 00:31:15 +00:00
|
|
|
{
|
|
|
|
__m128i r0 = _mm_load_si128(&src[i+0]);
|
|
|
|
__m128i r1 = _mm_load_si128(&src[i+1]);
|
|
|
|
__m128i r2 = _mm_load_si128(&src[i+2]);
|
|
|
|
__m128i r3 = _mm_load_si128(&src[i+3]);
|
|
|
|
|
|
|
|
_mm_store_si128(&dst[i*2+0], _mm_unpacklo_epi64(r0, r1));
|
|
|
|
_mm_store_si128(&dst[i*2+1], _mm_unpacklo_epi64(r2, r3));
|
|
|
|
_mm_store_si128(&dst[i*2+2], _mm_unpackhi_epi64(r0, r1));
|
|
|
|
_mm_store_si128(&dst[i*2+3], _mm_unpackhi_epi64(r2, r3));
|
|
|
|
|
|
|
|
__m128i r4 = _mm_load_si128(&src[i+0+16]);
|
|
|
|
__m128i r5 = _mm_load_si128(&src[i+1+16]);
|
|
|
|
__m128i r6 = _mm_load_si128(&src[i+2+16]);
|
|
|
|
__m128i r7 = _mm_load_si128(&src[i+3+16]);
|
|
|
|
|
|
|
|
_mm_store_si128(&dst[i*2+4], _mm_unpacklo_epi64(r4, r5));
|
|
|
|
_mm_store_si128(&dst[i*2+5], _mm_unpacklo_epi64(r6, r7));
|
|
|
|
_mm_store_si128(&dst[i*2+6], _mm_unpackhi_epi64(r4, r5));
|
|
|
|
_mm_store_si128(&dst[i*2+7], _mm_unpackhi_epi64(r6, r7));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
extern "C" void __fastcall WriteCLUT_T32_I4_CSM1_sse2(u32* vm, u32* clut)
|
|
|
|
{
|
|
|
|
__m128i* src = (__m128i*)vm;
|
|
|
|
__m128i* dst = (__m128i*)clut;
|
|
|
|
|
|
|
|
__m128i r0 = _mm_load_si128(&src[0]);
|
|
|
|
__m128i r1 = _mm_load_si128(&src[1]);
|
|
|
|
__m128i r2 = _mm_load_si128(&src[2]);
|
|
|
|
__m128i r3 = _mm_load_si128(&src[3]);
|
|
|
|
|
|
|
|
_mm_store_si128(&dst[0], _mm_unpacklo_epi64(r0, r1));
|
|
|
|
_mm_store_si128(&dst[1], _mm_unpacklo_epi64(r2, r3));
|
|
|
|
_mm_store_si128(&dst[2], _mm_unpackhi_epi64(r0, r1));
|
|
|
|
_mm_store_si128(&dst[3], _mm_unpackhi_epi64(r2, r3));
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
extern "C"
|
|
|
|
{
|
|
|
|
PCSX2_ALIGNED16(int s_clut16mask2[4]) = { 0x0000ffff, 0x0000ffff, 0x0000ffff, 0x0000ffff };
|
|
|
|
PCSX2_ALIGNED16(int s_clut16mask[8]) = { 0xffff0000, 0xffff0000, 0xffff0000, 0xffff0000,
|
|
|
|
0x0000ffff, 0x0000ffff, 0x0000ffff, 0x0000ffff
|
|
|
|
};
|
2010-03-19 00:31:15 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
extern "C" void __fastcall WriteCLUT_T16_I4_CSM1_sse2(u32* vm, u32* clut)
|
|
|
|
{
|
2010-04-01 22:52:59 +00:00
|
|
|
#if defined(_MSC_VER)
|
2010-05-01 20:33:53 +00:00
|
|
|
__asm
|
|
|
|
{
|
2010-03-19 00:31:15 +00:00
|
|
|
mov eax, vm
|
|
|
|
mov ecx, clut
|
|
|
|
movdqa xmm0, qword ptr [eax]
|
|
|
|
movdqa xmm1, qword ptr [eax+16]
|
|
|
|
movdqa xmm2, qword ptr [eax+32]
|
|
|
|
movdqa xmm3, qword ptr [eax+48]
|
|
|
|
|
|
|
|
// rearrange
|
|
|
|
pshuflw xmm0, xmm0, 0x88
|
|
|
|
pshufhw xmm0, xmm0, 0x88
|
|
|
|
pshuflw xmm1, xmm1, 0x88
|
|
|
|
pshufhw xmm1, xmm1, 0x88
|
|
|
|
pshuflw xmm2, xmm2, 0x88
|
|
|
|
pshufhw xmm2, xmm2, 0x88
|
|
|
|
pshuflw xmm3, xmm3, 0x88
|
|
|
|
pshufhw xmm3, xmm3, 0x88
|
|
|
|
|
|
|
|
shufps xmm0, xmm1, 0x88
|
|
|
|
shufps xmm2, xmm3, 0x88
|
|
|
|
|
|
|
|
pshufd xmm0, xmm0, 0xd8
|
|
|
|
pshufd xmm2, xmm2, 0xd8
|
|
|
|
|
|
|
|
pxor xmm6, xmm6
|
|
|
|
|
|
|
|
test ecx, 15
|
|
|
|
jnz WriteUnaligned
|
|
|
|
|
|
|
|
movdqa xmm7, s_clut16mask // saves upper 16 bits
|
|
|
|
|
|
|
|
// have to save interlaced with the old data
|
|
|
|
movdqa xmm4, [ecx]
|
|
|
|
movdqa xmm5, [ecx+32]
|
|
|
|
movhlps xmm1, xmm0
|
|
|
|
movlhps xmm0, xmm2 // lower 8 colors
|
|
|
|
|
|
|
|
pand xmm4, xmm7
|
|
|
|
pand xmm5, xmm7
|
|
|
|
|
|
|
|
shufps xmm1, xmm2, 0xe4 // upper 8 colors
|
|
|
|
movdqa xmm2, xmm0
|
|
|
|
movdqa xmm3, xmm1
|
|
|
|
|
|
|
|
punpcklwd xmm0, xmm6
|
|
|
|
punpcklwd xmm1, xmm6
|
|
|
|
por xmm0, xmm4
|
|
|
|
por xmm1, xmm5
|
|
|
|
|
|
|
|
punpckhwd xmm2, xmm6
|
|
|
|
punpckhwd xmm3, xmm6
|
|
|
|
|
|
|
|
movdqa [ecx], xmm0
|
|
|
|
movdqa [ecx+32], xmm1
|
|
|
|
|
|
|
|
movdqa xmm5, xmm7
|
|
|
|
pand xmm7, [ecx+16]
|
|
|
|
pand xmm5, [ecx+48]
|
|
|
|
|
|
|
|
por xmm2, xmm7
|
|
|
|
por xmm3, xmm5
|
|
|
|
|
|
|
|
movdqa [ecx+16], xmm2
|
|
|
|
movdqa [ecx+48], xmm3
|
|
|
|
jmp End
|
|
|
|
|
|
|
|
WriteUnaligned:
|
|
|
|
// ecx is offset by 2
|
|
|
|
sub ecx, 2
|
|
|
|
|
|
|
|
movdqa xmm7, s_clut16mask2 // saves lower 16 bits
|
|
|
|
|
|
|
|
// have to save interlaced with the old data
|
|
|
|
movdqa xmm4, [ecx]
|
|
|
|
movdqa xmm5, [ecx+32]
|
|
|
|
movhlps xmm1, xmm0
|
|
|
|
movlhps xmm0, xmm2 // lower 8 colors
|
|
|
|
|
|
|
|
pand xmm4, xmm7
|
|
|
|
pand xmm5, xmm7
|
|
|
|
|
|
|
|
shufps xmm1, xmm2, 0xe4 // upper 8 colors
|
|
|
|
movdqa xmm2, xmm0
|
|
|
|
movdqa xmm3, xmm1
|
|
|
|
|
|
|
|
punpcklwd xmm0, xmm6
|
|
|
|
punpcklwd xmm1, xmm6
|
|
|
|
pslld xmm0, 16
|
|
|
|
pslld xmm1, 16
|
|
|
|
por xmm0, xmm4
|
|
|
|
por xmm1, xmm5
|
|
|
|
|
|
|
|
punpckhwd xmm2, xmm6
|
|
|
|
punpckhwd xmm3, xmm6
|
|
|
|
pslld xmm2, 16
|
|
|
|
pslld xmm3, 16
|
|
|
|
|
|
|
|
movdqa [ecx], xmm0
|
|
|
|
movdqa [ecx+32], xmm1
|
|
|
|
|
|
|
|
movdqa xmm5, xmm7
|
|
|
|
pand xmm7, [ecx+16]
|
|
|
|
pand xmm5, [ecx+48]
|
|
|
|
|
|
|
|
por xmm2, xmm7
|
|
|
|
por xmm3, xmm5
|
|
|
|
|
|
|
|
movdqa [ecx+16], xmm2
|
|
|
|
movdqa [ecx+48], xmm3
|
2010-05-01 20:33:53 +00:00
|
|
|
|
2010-03-19 00:31:15 +00:00
|
|
|
End:
|
|
|
|
}
|
2010-04-01 22:52:59 +00:00
|
|
|
#else
|
|
|
|
__asm__(".intel_syntax noprefix\n"
|
2010-05-01 20:33:53 +00:00
|
|
|
"movdqa xmm0, xmmword ptr [ecx]\n"
|
|
|
|
"movdqa xmm1, xmmword ptr [ecx+16]\n"
|
|
|
|
"movdqa xmm2, xmmword ptr [ecx+32]\n"
|
|
|
|
"movdqa xmm3, xmmword ptr [ecx+48]\n"
|
2010-04-01 22:52:59 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
// rearrange
|
|
|
|
"pshuflw xmm0, xmm0, 0x88\n"
|
|
|
|
"pshufhw xmm0, xmm0, 0x88\n"
|
|
|
|
"pshuflw xmm1, xmm1, 0x88\n"
|
|
|
|
"pshufhw xmm1, xmm1, 0x88\n"
|
|
|
|
"pshuflw xmm2, xmm2, 0x88\n"
|
|
|
|
"pshufhw xmm2, xmm2, 0x88\n"
|
|
|
|
"pshuflw xmm3, xmm3, 0x88\n"
|
|
|
|
"pshufhw xmm3, xmm3, 0x88\n"
|
2010-04-01 22:52:59 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"shufps xmm0, xmm1, 0x88\n"
|
|
|
|
"shufps xmm2, xmm3, 0x88\n"
|
2010-04-01 22:52:59 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"pshufd xmm0, xmm0, 0xd8\n"
|
|
|
|
"pshufd xmm2, xmm2, 0xd8\n"
|
2010-04-01 22:52:59 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"pxor xmm6, xmm6\n"
|
2010-04-01 22:52:59 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"test edx, 15\n"
|
|
|
|
"jnz WriteUnaligned\n"
|
2010-04-01 22:52:59 +00:00
|
|
|
|
2010-06-21 12:03:14 +00:00
|
|
|
"movdqa xmm7, [s_clut16mask]\n" // saves upper 16 bits
|
2010-04-01 22:52:59 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
// have to save interlaced with the old data
|
|
|
|
"movdqa xmm4, [edx]\n"
|
|
|
|
"movdqa xmm5, [edx+32]\n"
|
|
|
|
"movhlps xmm1, xmm0\n"
|
|
|
|
"movlhps xmm0, xmm2\n"// lower 8 colors
|
2010-03-19 00:31:15 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"pand xmm4, xmm7\n"
|
|
|
|
"pand xmm5, xmm7\n"
|
2010-04-01 22:52:59 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"shufps xmm1, xmm2, 0xe4\n" // upper 8 colors
|
|
|
|
"movdqa xmm2, xmm0\n"
|
|
|
|
"movdqa xmm3, xmm1\n"
|
2010-04-01 22:52:59 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"punpcklwd xmm0, xmm6\n"
|
|
|
|
"punpcklwd xmm1, xmm6\n"
|
|
|
|
"por xmm0, xmm4\n"
|
|
|
|
"por xmm1, xmm5\n"
|
2010-04-01 22:52:59 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"punpckhwd xmm2, xmm6\n"
|
|
|
|
"punpckhwd xmm3, xmm6\n"
|
2010-04-01 22:52:59 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"movdqa [edx], xmm0\n"
|
|
|
|
"movdqa [edx+32], xmm1\n"
|
2010-04-01 22:52:59 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"movdqa xmm5, xmm7\n"
|
|
|
|
"pand xmm7, [edx+16]\n"
|
|
|
|
"pand xmm5, [edx+48]\n"
|
2010-04-01 22:52:59 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"por xmm2, xmm7\n"
|
|
|
|
"por xmm3, xmm5\n"
|
2010-04-01 22:52:59 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"movdqa [edx+16], xmm2\n"
|
|
|
|
"movdqa [edx+48], xmm3\n"
|
|
|
|
"jmp WriteCLUT_T16_I4_CSM1_End\n"
|
2010-04-01 22:52:59 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"WriteUnaligned:\n"
|
|
|
|
// %edx is offset by 2
|
|
|
|
"sub edx, 2\n"
|
2010-04-01 22:52:59 +00:00
|
|
|
|
2010-06-21 12:03:14 +00:00
|
|
|
"movdqa xmm7, [[s_clut16mask2]]\n" // saves lower 16 bits
|
2010-04-01 22:52:59 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
// have to save interlaced with the old data
|
|
|
|
"movdqa xmm4, [edx]\n"
|
|
|
|
"movdqa xmm5, [edx+32]\n"
|
|
|
|
"movhlps xmm1, xmm0\n"
|
|
|
|
"movlhps xmm0, xmm2\n" // lower 8 colors
|
|
|
|
|
|
|
|
"pand xmm4, xmm7\n"
|
|
|
|
"pand xmm5, xmm7\n"
|
|
|
|
|
|
|
|
"shufps xmm1, xmm2, 0xe4\n" // upper 8 colors
|
|
|
|
"movdqa xmm2, xmm0\n"
|
|
|
|
"movdqa xmm3, xmm1\n"
|
|
|
|
|
|
|
|
"punpcklwd xmm0, xmm6\n"
|
|
|
|
"punpcklwd xmm1, xmm6\n"
|
|
|
|
"pslld xmm0, 16\n"
|
|
|
|
"pslld xmm1, 16\n"
|
|
|
|
"por xmm0, xmm4\n"
|
|
|
|
"por xmm1, xmm5\n"
|
|
|
|
|
|
|
|
"punpckhwd xmm2, xmm6\n"
|
|
|
|
"punpckhwd xmm3, xmm6\n"
|
|
|
|
"pslld xmm2, 16\n"
|
|
|
|
"pslld xmm3, 16\n"
|
|
|
|
|
|
|
|
"movdqa [edx], xmm0\n"
|
|
|
|
"movdqa [edx+32], xmm1\n"
|
|
|
|
|
|
|
|
"movdqa xmm5, xmm7\n"
|
|
|
|
"pand xmm7, [edx+16]\n"
|
|
|
|
"pand xmm5, [edx+48]\n"
|
|
|
|
|
|
|
|
"por xmm2, xmm7\n"
|
|
|
|
"por xmm3, xmm5\n"
|
|
|
|
|
|
|
|
"movdqa [edx+16], xmm2\n"
|
|
|
|
"movdqa [edx+48], xmm3\n"
|
|
|
|
"WriteCLUT_T16_I4_CSM1_End:\n"
|
2010-06-21 12:03:14 +00:00
|
|
|
"\n"
|
|
|
|
".att_syntax\n" : [s_clut16mask] "=m" (s_clut16mask), [s_clut16mask2] "=m" (s_clut16mask2)
|
2010-05-01 20:33:53 +00:00
|
|
|
);
|
2010-04-01 22:52:59 +00:00
|
|
|
#endif // _MSC_VER
|
|
|
|
}
|
2010-05-01 20:33:53 +00:00
|
|
|
|
2010-03-19 00:31:15 +00:00
|
|
|
#endif // ZEROGS_SSE2
|
|
|
|
|
|
|
|
void __fastcall WriteCLUT_T16_I8_CSM1_c(u32* _vm, u32* _clut)
|
|
|
|
{
|
2010-04-25 00:31:27 +00:00
|
|
|
const static u32 map[] =
|
2010-03-19 00:31:15 +00:00
|
|
|
{
|
|
|
|
0, 2, 8, 10, 16, 18, 24, 26,
|
|
|
|
4, 6, 12, 14, 20, 22, 28, 30,
|
2010-04-25 00:31:27 +00:00
|
|
|
1, 3, 9, 11, 17, 19, 25, 27,
|
2010-03-19 00:31:15 +00:00
|
|
|
5, 7, 13, 15, 21, 23, 29, 31
|
|
|
|
};
|
|
|
|
|
|
|
|
u16* vm = (u16*)_vm;
|
|
|
|
u16* clut = (u16*)_clut;
|
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
int left = ((u32)(uptr)clut & 2) ? 512 : 512 - (((u32)(uptr)clut) & 0x3ff) / 2;
|
2010-03-19 00:31:15 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
for (int j = 0; j < 8; j++, vm += 32, clut += 64, left -= 32)
|
2010-03-19 00:31:15 +00:00
|
|
|
{
|
2010-05-01 20:33:53 +00:00
|
|
|
if (left == 32)
|
|
|
|
{
|
|
|
|
assert(left == 32);
|
|
|
|
|
|
|
|
for (int i = 0; i < 16; i++)
|
2010-03-19 00:31:15 +00:00
|
|
|
clut[2*i] = vm[map[i]];
|
|
|
|
|
|
|
|
clut = (u16*)((uptr)clut & ~0x3ff) + 1;
|
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
for (int i = 16; i < 32; i++)
|
2010-03-19 00:31:15 +00:00
|
|
|
clut[2*i] = vm[map[i]];
|
|
|
|
}
|
2010-05-01 20:33:53 +00:00
|
|
|
else
|
|
|
|
{
|
|
|
|
if (left == 0)
|
|
|
|
{
|
2010-03-19 00:31:15 +00:00
|
|
|
clut = (u16*)((uptr)clut & ~0x3ff) + 1;
|
|
|
|
left = -1;
|
|
|
|
}
|
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
for (int i = 0; i < 32; i++)
|
2010-03-19 00:31:15 +00:00
|
|
|
clut[2*i] = vm[map[i]];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void __fastcall WriteCLUT_T32_I8_CSM1_c(u32* vm, u32* clut)
|
|
|
|
{
|
|
|
|
u64* src = (u64*)vm;
|
|
|
|
u64* dst = (u64*)clut;
|
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
for (int j = 0; j < 2; j++, src += 32)
|
|
|
|
{
|
|
|
|
for (int i = 0; i < 4; i++, dst += 16, src += 8)
|
2010-03-19 00:31:15 +00:00
|
|
|
{
|
|
|
|
dst[0] = src[0];
|
|
|
|
dst[1] = src[2];
|
|
|
|
dst[2] = src[4];
|
|
|
|
dst[3] = src[6];
|
|
|
|
dst[4] = src[1];
|
|
|
|
dst[5] = src[3];
|
|
|
|
dst[6] = src[5];
|
|
|
|
dst[7] = src[7];
|
|
|
|
|
|
|
|
dst[8] = src[32];
|
|
|
|
dst[9] = src[32+2];
|
|
|
|
dst[10] = src[32+4];
|
|
|
|
dst[11] = src[32+6];
|
|
|
|
dst[12] = src[32+1];
|
|
|
|
dst[13] = src[32+3];
|
|
|
|
dst[14] = src[32+5];
|
|
|
|
dst[15] = src[32+7];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void __fastcall WriteCLUT_T16_I4_CSM1_c(u32* _vm, u32* _clut)
|
|
|
|
{
|
|
|
|
u16* dst = (u16*)_clut;
|
|
|
|
u16* src = (u16*)_vm;
|
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
dst[0] = src[0];
|
|
|
|
dst[2] = src[2];
|
|
|
|
dst[4] = src[8];
|
|
|
|
dst[6] = src[10];
|
|
|
|
dst[8] = src[16];
|
|
|
|
dst[10] = src[18];
|
|
|
|
dst[12] = src[24];
|
|
|
|
dst[14] = src[26];
|
|
|
|
dst[16] = src[4];
|
|
|
|
dst[18] = src[6];
|
|
|
|
dst[20] = src[12];
|
|
|
|
dst[22] = src[14];
|
|
|
|
dst[24] = src[20];
|
|
|
|
dst[26] = src[22];
|
|
|
|
dst[28] = src[28];
|
|
|
|
dst[30] = src[30];
|
2010-03-19 00:31:15 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
void __fastcall WriteCLUT_T32_I4_CSM1_c(u32* vm, u32* clut)
|
|
|
|
{
|
|
|
|
u64* src = (u64*)vm;
|
|
|
|
u64* dst = (u64*)clut;
|
|
|
|
|
|
|
|
dst[0] = src[0];
|
|
|
|
dst[1] = src[2];
|
|
|
|
dst[2] = src[4];
|
|
|
|
dst[3] = src[6];
|
|
|
|
dst[4] = src[1];
|
|
|
|
dst[5] = src[3];
|
|
|
|
dst[6] = src[5];
|
|
|
|
dst[7] = src[7];
|
|
|
|
}
|
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
void SSE2_UnswizzleZ16Target(u16* dst, u16* src, int iters)
|
|
|
|
{
|
2010-03-19 00:31:15 +00:00
|
|
|
|
2010-04-25 00:31:27 +00:00
|
|
|
#if defined(_MSC_VER)
|
2010-05-01 20:33:53 +00:00
|
|
|
__asm
|
|
|
|
{
|
2010-03-19 00:31:15 +00:00
|
|
|
mov edx, iters
|
|
|
|
pxor xmm7, xmm7
|
|
|
|
mov eax, dst
|
|
|
|
mov ecx, src
|
|
|
|
|
|
|
|
Z16Loop:
|
|
|
|
// unpack 64 bytes at a time
|
|
|
|
movdqa xmm0, [ecx]
|
|
|
|
movdqa xmm2, [ecx+16]
|
|
|
|
movdqa xmm4, [ecx+32]
|
|
|
|
movdqa xmm6, [ecx+48]
|
|
|
|
|
|
|
|
movdqa xmm1, xmm0
|
|
|
|
movdqa xmm3, xmm2
|
|
|
|
movdqa xmm5, xmm4
|
|
|
|
|
|
|
|
punpcklwd xmm0, xmm7
|
|
|
|
punpckhwd xmm1, xmm7
|
|
|
|
punpcklwd xmm2, xmm7
|
|
|
|
punpckhwd xmm3, xmm7
|
|
|
|
|
|
|
|
// start saving
|
|
|
|
movdqa [eax], xmm0
|
|
|
|
movdqa [eax+16], xmm1
|
|
|
|
|
|
|
|
punpcklwd xmm4, xmm7
|
|
|
|
punpckhwd xmm5, xmm7
|
|
|
|
|
|
|
|
movdqa [eax+32], xmm2
|
|
|
|
movdqa [eax+48], xmm3
|
|
|
|
|
|
|
|
movdqa xmm0, xmm6
|
|
|
|
punpcklwd xmm6, xmm7
|
|
|
|
|
|
|
|
movdqa [eax+64], xmm4
|
|
|
|
movdqa [eax+80], xmm5
|
|
|
|
|
|
|
|
punpckhwd xmm0, xmm7
|
|
|
|
|
|
|
|
movdqa [eax+96], xmm6
|
|
|
|
movdqa [eax+112], xmm0
|
|
|
|
|
|
|
|
add ecx, 64
|
|
|
|
add eax, 128
|
|
|
|
sub edx, 1
|
|
|
|
jne Z16Loop
|
|
|
|
}
|
|
|
|
#else // _MSC_VER
|
|
|
|
|
|
|
|
__asm__(".intel_syntax\n"
|
2010-05-01 20:33:53 +00:00
|
|
|
"pxor %%xmm7, %%xmm7\n"
|
2010-04-25 00:31:27 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"Z16Loop:\n"
|
|
|
|
// unpack 64 bytes at a time
|
|
|
|
"movdqa %%xmm0, [%0]\n"
|
|
|
|
"movdqa %%xmm2, [%0+16]\n"
|
|
|
|
"movdqa %%xmm4, [%0+32]\n"
|
|
|
|
"movdqa %%xmm6, [%0+48]\n"
|
2010-04-25 00:31:27 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"movdqa %%xmm1, %%xmm0\n"
|
|
|
|
"movdqa %%xmm3, %%xmm2\n"
|
|
|
|
"movdqa %%xmm5, %%xmm4\n"
|
2010-04-25 00:31:27 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"punpcklwd %%xmm0, %%xmm7\n"
|
|
|
|
"punpckhwd %%xmm1, %%xmm7\n"
|
|
|
|
"punpcklwd %%xmm2, %%xmm7\n"
|
|
|
|
"punpckhwd %%xmm3, %%xmm7\n"
|
2010-04-25 00:31:27 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
// start saving
|
|
|
|
"movdqa [%1], %%xmm0\n"
|
|
|
|
"movdqa [%1+16], %%xmm1\n"
|
|
|
|
|
|
|
|
"punpcklwd %%xmm4, %%xmm7\n"
|
|
|
|
"punpckhwd %%xmm5, %%xmm7\n"
|
2010-04-25 00:31:27 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"movdqa [%1+32], %%xmm2\n"
|
|
|
|
"movdqa [%1+48], %%xmm3\n"
|
2010-04-25 00:31:27 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"movdqa %%xmm0, %%xmm6\n"
|
|
|
|
"punpcklwd %%xmm6, %%xmm7\n"
|
2010-04-25 00:31:27 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"movdqa [%1+64], %%xmm4\n"
|
|
|
|
"movdqa [%1+80], %%xmm5\n"
|
2010-04-25 00:31:27 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"punpckhwd %%xmm0, %%xmm7\n"
|
2010-03-19 00:31:15 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"movdqa [%1+96], %%xmm6\n"
|
|
|
|
"movdqa [%1+112], %%xmm0\n"
|
2010-04-25 00:31:27 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
"add %0, 64\n"
|
|
|
|
"add %1, 128\n"
|
|
|
|
"sub %2, 1\n"
|
|
|
|
"jne Z16Loop\n"
|
2010-04-25 00:31:27 +00:00
|
|
|
|
2010-05-01 20:33:53 +00:00
|
|
|
".att_syntax\n" : "=r"(src), "=r"(dst), "=r"(iters) : "0"(src), "1"(dst), "2"(iters)
|
|
|
|
);
|
2010-03-19 00:31:15 +00:00
|
|
|
#endif // _MSC_VER
|
|
|
|
}
|
|
|
|
|