| Commit message (Collapse) | Author | Age | Files | Lines | |
|---|---|---|---|---|---|
| * | SBR DSP x86: implement SSE sbr_hf_g_filt | Christophe GISQUET | 2012-02-23 | 1 | -0/+3 |
| | | | | | | | | | | | | | | | | Unrolling the main loop to process, instead of 4 elements: - 8: minor gain of 2 cycles (not worth the extra object size) - 2: loss of 8 cycles. Assigning STEP to a register is a loss. Output address (Y) is almost always unaligned. Timings: - C (32/64 bits): 117/109 cycles - SSE: 57 cycles Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com> | ||||
| * | SBR DSP x86: implement SSE sbr_sum_square_sse | Christophe GISQUET | 2012-02-23 | 1 | -0/+37 |
| The 32bits targets have been compiled with -mfpmath=sse for proper reference. sbr_sum_square C /32bits: 82c (unrolled)/102c C /64bits: 69c (unrolled)/82c SSE/32bits: 42c SSE/64bits: 31c Use of SSE4.1 dpps to perform the final sum is slower. Not unrolling to perform 8 operations in a loop yields 10 more cycles. Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com> | |||||
 |
index : ffmpeg | |
| Mirror of FFmpeg git repo |
| aboutsummaryrefslogtreecommitdiffstats |