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
Even if we're blitting between two different surfaces their pixels might still overlap, because of SDL_CreateRGBSurfaceFrom(), so always use SDL_BlitCopy() and check for overlap in that function.
When handling overlapping surfaces, don't assume that memcpy() iterates forward, instead use memmove() correctly, and provide a fallback implementation of SDL_memmove() that handles the different cases.
Fixed a bug with SDL_memset() not completely filling lengths that aren't a multiple of 4.
Optimized SDL_memcpy() a bit using the same technique as SDL_memset().
Don't use the SSE cache instruction in MMX code if SSE isn't available. :)
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%404326
Moved fill and copy routines to their own files.
--HG--
rename : src/video/SDL_blit_copy.c => src/video/SDL_copy.c
rename : src/video/SDL_blit_copy.h => src/video/SDL_copy.h
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%402615
Added SDL_blit_copy.c to the Visual C++ project
The SSE and MMX intrinsics don't compile on Visual Studio yet...
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%402612
Prepping for using MMX and SSE intrinsics instead of inline assembly.
.. except for memcpy equivalents which only get faster if they can
exploit the parallelism of loading into multiple SIMD registers. :)
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%402609
