diff options
| author | Justin Ruggles <justin.ruggles@gmail.com> | 2011-10-09 23:52:03 -0400 |
|---|---|---|
| committer | Justin Ruggles <justin.ruggles@gmail.com> | 2011-10-21 10:13:05 -0400 |
| commit | 4e8e2624767f4af0eaa932c543d072fed96fd586 (patch) | |
| tree | 56c363bfc5cdff42c6346d883680d620826fde22 /libavcodec/x86/fmtconvert.asm | |
| parent | 185142a5ea93ef723f70a3ea43797f6c8827eb79 (diff) | |
fmtconvert: port int32_to_float_fmul_scalar() x86 inline asm to yasm
Diffstat (limited to 'libavcodec/x86/fmtconvert.asm')
| -rw-r--r-- | libavcodec/x86/fmtconvert.asm | 46 |
1 files changed, 46 insertions, 0 deletions
diff --git a/libavcodec/x86/fmtconvert.asm b/libavcodec/x86/fmtconvert.asm index d314a4e14e..e3eb5d2286 100644 --- a/libavcodec/x86/fmtconvert.asm +++ b/libavcodec/x86/fmtconvert.asm @@ -24,6 +24,52 @@ SECTION_TEXT +;--------------------------------------------------------------------------------- +; void int32_to_float_fmul_scalar(float *dst, const int *src, float mul, int len); +;--------------------------------------------------------------------------------- +%macro INT32_TO_FLOAT_FMUL_SCALAR 2 +%ifdef ARCH_X86_64 +cglobal int32_to_float_fmul_scalar_%1, 3,3,%2, dst, src, len +%else +cglobal int32_to_float_fmul_scalar_%1, 4,4,%2, dst, src, mul, len + movss m0, mulm +%endif + SPLATD m0 + shl lenq, 2 + add srcq, lenq + add dstq, lenq + neg lenq +.loop: +%ifidn %1, sse2 + cvtdq2ps m1, [srcq+lenq ] + cvtdq2ps m2, [srcq+lenq+16] +%else + cvtpi2ps m1, [srcq+lenq ] + cvtpi2ps m3, [srcq+lenq+ 8] + cvtpi2ps m2, [srcq+lenq+16] + cvtpi2ps m4, [srcq+lenq+24] + movlhps m1, m3 + movlhps m2, m4 +%endif + mulps m1, m0 + mulps m2, m0 + mova [dstq+lenq ], m1 + mova [dstq+lenq+16], m2 + add lenq, 32 + jl .loop + REP_RET +%endmacro + +INIT_XMM +%define SPLATD SPLATD_SSE +%define movdqa movaps +INT32_TO_FLOAT_FMUL_SCALAR sse, 5 +%undef movdqa +%define SPLATD SPLATD_SSE2 +INT32_TO_FLOAT_FMUL_SCALAR sse2, 3 +%undef SPLATD + + ;------------------------------------------------------------------------------ ; void ff_float_to_int16(int16_t *dst, const float *src, long len); ;------------------------------------------------------------------------------ |