diff options
author | Christophe Gisquet <christophe.gisquet@gmail.com> | 2012-11-25 02:15:07 +0400 |
---|---|---|
committer | Diego Biurrun <diego@biurrun.de> | 2013-01-06 16:57:01 +0400 |
commit | 44a0036d10579ed91e48df24859e54b08a582742 (patch) | |
tree | 9efe5bd75b5db9a0fc2b921d2fd593129102d4ec /libavcodec/x86/sbrdsp.asm | |
parent | 5b4dfbffc258f90a7d2540d21209ac23afcf7cd0 (diff) |
x86: sbrdsp: Implement SSE sum64x5
698 to 174 cycles on Arrandale. Unrolling is a 6 cycles gain.
Signed-off-by: Diego Biurrun <diego@biurrun.de>
Diffstat (limited to 'libavcodec/x86/sbrdsp.asm')
-rw-r--r-- | libavcodec/x86/sbrdsp.asm | 22 |
1 files changed, 22 insertions, 0 deletions
diff --git a/libavcodec/x86/sbrdsp.asm b/libavcodec/x86/sbrdsp.asm index b87da4a072..063c24f95c 100644 --- a/libavcodec/x86/sbrdsp.asm +++ b/libavcodec/x86/sbrdsp.asm @@ -181,3 +181,25 @@ cglobal sbr_hf_gen, 4,4,8, X_high, X_low, alpha0, alpha1, BW, S, E add start, 16 jnz .loop2 RET + +cglobal sbr_sum64x5, 1,2,4,z + lea r1q, [zq+ 256] +.loop: + mova m0, [zq+ 0] + mova m2, [zq+ 16] + mova m1, [zq+ 256] + mova m3, [zq+ 272] + addps m0, [zq+ 512] + addps m2, [zq+ 528] + addps m1, [zq+ 768] + addps m3, [zq+ 784] + addps m0, [zq+1024] + addps m2, [zq+1040] + addps m0, m1 + addps m2, m3 + mova [zq], m0 + mova [zq+16], m2 + add zq, 32 + cmp zq, r1q + jne .loop + REP_RET |