Mercurial > libavcodec.hg
comparison x86/dsputil_mmx.c @ 10301:02798c603744 libavcodec
cosmetics: fix indentation after previous commit
| author | mru |
|---|---|
| date | Sun, 27 Sep 2009 16:52:00 +0000 |
| parents | 4d1b9ca628fc |
| children | 12c8175d6db5 |
comparison
equal
deleted
inserted
replaced
| 10300:4d1b9ca628fc | 10301:02798c603744 |
|---|---|
| 2126 } | 2126 } |
| 2127 | 2127 |
| 2128 static void vector_fmul_add_3dnow(float *dst, const float *src0, const float *src1, | 2128 static void vector_fmul_add_3dnow(float *dst, const float *src0, const float *src1, |
| 2129 const float *src2, int len){ | 2129 const float *src2, int len){ |
| 2130 x86_reg i = (len-4)*4; | 2130 x86_reg i = (len-4)*4; |
| 2131 __asm__ volatile( | 2131 __asm__ volatile( |
| 2132 "1: \n\t" | 2132 "1: \n\t" |
| 2133 "movq (%2,%0), %%mm0 \n\t" | 2133 "movq (%2,%0), %%mm0 \n\t" |
| 2134 "movq 8(%2,%0), %%mm1 \n\t" | 2134 "movq 8(%2,%0), %%mm1 \n\t" |
| 2135 "pfmul (%3,%0), %%mm0 \n\t" | 2135 "pfmul (%3,%0), %%mm0 \n\t" |
| 2136 "pfmul 8(%3,%0), %%mm1 \n\t" | 2136 "pfmul 8(%3,%0), %%mm1 \n\t" |
| 2137 "pfadd (%4,%0), %%mm0 \n\t" | 2137 "pfadd (%4,%0), %%mm0 \n\t" |
| 2138 "pfadd 8(%4,%0), %%mm1 \n\t" | 2138 "pfadd 8(%4,%0), %%mm1 \n\t" |
| 2139 "movq %%mm0, (%1,%0) \n\t" | 2139 "movq %%mm0, (%1,%0) \n\t" |
| 2140 "movq %%mm1, 8(%1,%0) \n\t" | 2140 "movq %%mm1, 8(%1,%0) \n\t" |
| 2141 "sub $16, %0 \n\t" | 2141 "sub $16, %0 \n\t" |
| 2142 "jge 1b \n\t" | 2142 "jge 1b \n\t" |
| 2143 :"+r"(i) | 2143 :"+r"(i) |
| 2144 :"r"(dst), "r"(src0), "r"(src1), "r"(src2) | 2144 :"r"(dst), "r"(src0), "r"(src1), "r"(src2) |
| 2145 :"memory" | 2145 :"memory" |
| 2146 ); | 2146 ); |
| 2147 __asm__ volatile("femms"); | 2147 __asm__ volatile("femms"); |
| 2148 } | 2148 } |
| 2149 static void vector_fmul_add_sse(float *dst, const float *src0, const float *src1, | 2149 static void vector_fmul_add_sse(float *dst, const float *src0, const float *src1, |
| 2150 const float *src2, int len){ | 2150 const float *src2, int len){ |
| 2151 x86_reg i = (len-8)*4; | 2151 x86_reg i = (len-8)*4; |
| 2152 __asm__ volatile( | 2152 __asm__ volatile( |
| 2153 "1: \n\t" | 2153 "1: \n\t" |
| 2154 "movaps (%2,%0), %%xmm0 \n\t" | 2154 "movaps (%2,%0), %%xmm0 \n\t" |
| 2155 "movaps 16(%2,%0), %%xmm1 \n\t" | 2155 "movaps 16(%2,%0), %%xmm1 \n\t" |
| 2156 "mulps (%3,%0), %%xmm0 \n\t" | 2156 "mulps (%3,%0), %%xmm0 \n\t" |
| 2157 "mulps 16(%3,%0), %%xmm1 \n\t" | 2157 "mulps 16(%3,%0), %%xmm1 \n\t" |
| 2158 "addps (%4,%0), %%xmm0 \n\t" | 2158 "addps (%4,%0), %%xmm0 \n\t" |
| 2159 "addps 16(%4,%0), %%xmm1 \n\t" | 2159 "addps 16(%4,%0), %%xmm1 \n\t" |
| 2160 "movaps %%xmm0, (%1,%0) \n\t" | 2160 "movaps %%xmm0, (%1,%0) \n\t" |
| 2161 "movaps %%xmm1, 16(%1,%0) \n\t" | 2161 "movaps %%xmm1, 16(%1,%0) \n\t" |
| 2162 "sub $32, %0 \n\t" | 2162 "sub $32, %0 \n\t" |
| 2163 "jge 1b \n\t" | 2163 "jge 1b \n\t" |
| 2164 :"+r"(i) | 2164 :"+r"(i) |
| 2165 :"r"(dst), "r"(src0), "r"(src1), "r"(src2) | 2165 :"r"(dst), "r"(src0), "r"(src1), "r"(src2) |
| 2166 :"memory" | 2166 :"memory" |
| 2167 ); | 2167 ); |
| 2168 } | 2168 } |
| 2169 | 2169 |
| 2170 static void vector_fmul_window_3dnow2(float *dst, const float *src0, const float *src1, | 2170 static void vector_fmul_window_3dnow2(float *dst, const float *src0, const float *src1, |
| 2171 const float *win, float add_bias, int len){ | 2171 const float *win, float add_bias, int len){ |
| 2172 #if HAVE_6REGS | 2172 #if HAVE_6REGS |