2007-08-15 08:21:10 +00:00
|
|
|
/*
|
2011-04-08 13:03:26 -07:00
|
|
|
Simple DirectMedia Layer
|
|
|
|
Copyright (C) 1997-2011 Sam Lantinga <slouken@libsdl.org>
|
|
|
|
|
|
|
|
This software is provided 'as-is', without any express or implied
|
|
|
|
warranty. In no event will the authors be held liable for any damages
|
|
|
|
arising from the use of this software.
|
|
|
|
|
|
|
|
Permission is granted to anyone to use this software for any purpose,
|
|
|
|
including commercial applications, and to alter it and redistribute it
|
|
|
|
freely, subject to the following restrictions:
|
|
|
|
|
|
|
|
1. The origin of this software must not be misrepresented; you must not
|
|
|
|
claim that you wrote the original software. If you use this software
|
|
|
|
in a product, an acknowledgment in the product documentation would be
|
|
|
|
appreciated but is not required.
|
|
|
|
2. Altered source versions must be plainly marked as such, and must not be
|
|
|
|
misrepresented as being the original software.
|
|
|
|
3. This notice may not be removed or altered from any source distribution.
|
2007-08-15 08:21:10 +00:00
|
|
|
*/
|
|
|
|
#include "SDL_config.h"
|
|
|
|
|
|
|
|
#include "SDL_video.h"
|
|
|
|
#include "SDL_blit.h"
|
2007-08-16 21:54:26 +00:00
|
|
|
#include "SDL_blit_copy.h"
|
2007-08-15 08:21:10 +00:00
|
|
|
|
|
|
|
|
2007-08-16 21:43:19 +00:00
|
|
|
#ifdef __SSE__
|
|
|
|
/* This assumes 16-byte aligned src and dst */
|
|
|
|
static __inline__ void
|
|
|
|
SDL_memcpySSE(Uint8 * dst, const Uint8 * src, int len)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
__m128 values[4];
|
|
|
|
for (i = len / 64; i--;) {
|
|
|
|
_mm_prefetch(src, _MM_HINT_NTA);
|
|
|
|
values[0] = *(__m128 *) (src + 0);
|
|
|
|
values[1] = *(__m128 *) (src + 16);
|
|
|
|
values[2] = *(__m128 *) (src + 32);
|
|
|
|
values[3] = *(__m128 *) (src + 48);
|
|
|
|
_mm_stream_ps((float *) (dst + 0), values[0]);
|
|
|
|
_mm_stream_ps((float *) (dst + 16), values[1]);
|
|
|
|
_mm_stream_ps((float *) (dst + 32), values[2]);
|
|
|
|
_mm_stream_ps((float *) (dst + 48), values[3]);
|
|
|
|
src += 64;
|
|
|
|
dst += 64;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (len & 63)
|
|
|
|
SDL_memcpy(dst, src, len & 63);
|
|
|
|
}
|
|
|
|
#endif /* __SSE__ */
|
|
|
|
|
2007-08-15 08:21:10 +00:00
|
|
|
#ifdef __MMX__
|
2008-11-24 21:43:02 +00:00
|
|
|
#ifdef _MSC_VER
|
|
|
|
#pragma warning(disable:4799)
|
|
|
|
#endif
|
2007-08-15 08:21:10 +00:00
|
|
|
static __inline__ void
|
2007-08-16 02:14:13 +00:00
|
|
|
SDL_memcpyMMX(Uint8 * dst, const Uint8 * src, int len)
|
2007-08-15 08:21:10 +00:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
Some MMX fixes from Patrick Baggett.
Original email...
Date: Sat, 10 Sep 2011 13:01:20 -0500
From: Patrick Baggett
To: SDL Development List <sdl@lists.libsdl.org>
Subject: Re: [SDL] SDL_memcpyMMX uses SSE instructions
In SDL_blit_copy.c, the function SDL_memcpyMMX() actually use SSE
instructions.
It is called in this context:
#ifdef __MMX__
if (SDL_HasMMX() &&
!((uintptr_t) src & 7) && !(srcskip & 7) &&
!((uintptr_t) dst & 7) && !(dstskip & 7)) {
while (h--) {
SDL_memcpyMMX(dst, src, w);
src += srcskip;
dst += dstskip;
}
_mm_empty();
return;
}
#endif
This implies that the minimum CPU features are just MMX. There is a
separate SDL_memcpySSE() function.
The SDL_memcpyMMX() function does:
#ifdef __SSE__
_mm_prefetch(src, _MM_HINT_NTA);
#endif
...which tests at compile time if SSE intrinsics are available, not at run
time. It generates the PREFETCHNTA instruction. It also uses _mm_stream_pi()
intrinsic, which generates the MOVNTQ instruction.
If you replace the "MMX" code with:
__m64* d64 = (__m64*)dst;
__m64* s64 = (__m64*)src;
for(i= len / 64; i--;) {
d64[0] = s64[0];
d64[1] = s64[1];
d64[2] = s64[2];
d64[3] = s64[3];
d64[4] = s64[4];
d64[5] = s64[5];
d64[6] = s64[6];
d64[7] = s64[7];
d64 += 8;
s64 += 8;
}
Then MSVC generates the correct movq instructions. GCC (4.5.0) seems to
think that using 2x movl is still better, but then again, GCC isn't actually
that good at optimizing intrinsics as I've found. At least the code won't
crash on my P2 though. :)
Also, there is no requirement for MMX to be aligned to the 8th byte. I
think the author assumed that SSE's 16 byte alignment requirement must
retroactively mean that MMX requires 8 byte alignment. Attached is the full
patch.
Patrick
2011-09-11 01:54:54 -04:00
|
|
|
__m64* d64 = (__m64*)dst;
|
|
|
|
__m64* s64 = (__m64*)src;
|
|
|
|
|
|
|
|
for(i= len / 64; i--;) {
|
|
|
|
|
|
|
|
d64[0] = s64[0];
|
|
|
|
d64[1] = s64[1];
|
|
|
|
d64[2] = s64[2];
|
|
|
|
d64[3] = s64[3];
|
|
|
|
d64[4] = s64[4];
|
|
|
|
d64[5] = s64[5];
|
|
|
|
d64[6] = s64[6];
|
|
|
|
d64[7] = s64[7];
|
|
|
|
|
|
|
|
d64 += 8;
|
|
|
|
s64 += 8;
|
2007-08-15 08:21:10 +00:00
|
|
|
}
|
2007-08-16 02:14:13 +00:00
|
|
|
|
2007-08-15 08:21:10 +00:00
|
|
|
if (len & 63)
|
|
|
|
SDL_memcpy(dst, src, len & 63);
|
|
|
|
}
|
|
|
|
#endif /* __MMX__ */
|
|
|
|
|
|
|
|
void
|
|
|
|
SDL_BlitCopy(SDL_BlitInfo * info)
|
|
|
|
{
|
2011-02-16 15:25:10 -08:00
|
|
|
SDL_bool overlap;
|
2007-08-15 08:21:10 +00:00
|
|
|
Uint8 *src, *dst;
|
|
|
|
int w, h;
|
|
|
|
int srcskip, dstskip;
|
|
|
|
|
2007-08-18 05:39:09 +00:00
|
|
|
w = info->dst_w * info->dst_fmt->BytesPerPixel;
|
2007-08-17 06:21:58 +00:00
|
|
|
h = info->dst_h;
|
|
|
|
src = info->src;
|
|
|
|
dst = info->dst;
|
2007-08-18 05:39:09 +00:00
|
|
|
srcskip = info->src_pitch;
|
|
|
|
dstskip = info->dst_pitch;
|
2007-08-15 08:21:10 +00:00
|
|
|
|
2011-02-16 15:25:10 -08:00
|
|
|
/* Properly handle overlapping blits */
|
|
|
|
if (src < dst) {
|
|
|
|
overlap = (dst < (src + h*srcskip));
|
|
|
|
} else {
|
|
|
|
overlap = (src < (dst + h*dstskip));
|
|
|
|
}
|
|
|
|
if (overlap) {
|
|
|
|
while (h--) {
|
|
|
|
SDL_memmove(dst, src, w);
|
|
|
|
src += srcskip;
|
|
|
|
dst += dstskip;
|
|
|
|
}
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2007-08-15 08:21:10 +00:00
|
|
|
#ifdef __SSE__
|
2008-12-03 06:32:04 +00:00
|
|
|
if (SDL_HasSSE() &&
|
|
|
|
!((uintptr_t) src & 15) && !(srcskip & 15) &&
|
|
|
|
!((uintptr_t) dst & 15) && !(dstskip & 15)) {
|
2007-08-15 08:21:10 +00:00
|
|
|
while (h--) {
|
|
|
|
SDL_memcpySSE(dst, src, w);
|
|
|
|
src += srcskip;
|
|
|
|
dst += dstskip;
|
|
|
|
}
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#ifdef __MMX__
|
Some MMX fixes from Patrick Baggett.
Original email...
Date: Sat, 10 Sep 2011 13:01:20 -0500
From: Patrick Baggett
To: SDL Development List <sdl@lists.libsdl.org>
Subject: Re: [SDL] SDL_memcpyMMX uses SSE instructions
In SDL_blit_copy.c, the function SDL_memcpyMMX() actually use SSE
instructions.
It is called in this context:
#ifdef __MMX__
if (SDL_HasMMX() &&
!((uintptr_t) src & 7) && !(srcskip & 7) &&
!((uintptr_t) dst & 7) && !(dstskip & 7)) {
while (h--) {
SDL_memcpyMMX(dst, src, w);
src += srcskip;
dst += dstskip;
}
_mm_empty();
return;
}
#endif
This implies that the minimum CPU features are just MMX. There is a
separate SDL_memcpySSE() function.
The SDL_memcpyMMX() function does:
#ifdef __SSE__
_mm_prefetch(src, _MM_HINT_NTA);
#endif
...which tests at compile time if SSE intrinsics are available, not at run
time. It generates the PREFETCHNTA instruction. It also uses _mm_stream_pi()
intrinsic, which generates the MOVNTQ instruction.
If you replace the "MMX" code with:
__m64* d64 = (__m64*)dst;
__m64* s64 = (__m64*)src;
for(i= len / 64; i--;) {
d64[0] = s64[0];
d64[1] = s64[1];
d64[2] = s64[2];
d64[3] = s64[3];
d64[4] = s64[4];
d64[5] = s64[5];
d64[6] = s64[6];
d64[7] = s64[7];
d64 += 8;
s64 += 8;
}
Then MSVC generates the correct movq instructions. GCC (4.5.0) seems to
think that using 2x movl is still better, but then again, GCC isn't actually
that good at optimizing intrinsics as I've found. At least the code won't
crash on my P2 though. :)
Also, there is no requirement for MMX to be aligned to the 8th byte. I
think the author assumed that SSE's 16 byte alignment requirement must
retroactively mean that MMX requires 8 byte alignment. Attached is the full
patch.
Patrick
2011-09-11 01:54:54 -04:00
|
|
|
if (SDL_HasMMX()) {
|
2007-08-15 08:21:10 +00:00
|
|
|
while (h--) {
|
|
|
|
SDL_memcpyMMX(dst, src, w);
|
|
|
|
src += srcskip;
|
|
|
|
dst += dstskip;
|
|
|
|
}
|
2007-08-16 02:14:13 +00:00
|
|
|
_mm_empty();
|
2007-08-15 08:21:10 +00:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
while (h--) {
|
|
|
|
SDL_memcpy(dst, src, w);
|
|
|
|
src += srcskip;
|
|
|
|
dst += dstskip;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* vi: set ts=4 sw=4 expandtab: */
|