summaryrefslogtreecommitdiffstats
path: root/libavcodec/sbrdsp.c
Commit message (Collapse)AuthorAgeFilesLines
* sbrdsp: unroll and use integer operationsChristophe Gisquet2013-04-061-12/+28
| | | | | | | | | | | | This patch can be controversial, by assuming floats are IEEE-754 and particular behaviour of the FPU will get in the way. Timing on Arrandale and Win32 (thus, x87 FPU is used in the reference). sbr_qmf_pre_shuffle_c: 115 to 76 sbr_neg_odd_64_c: 84 to 55 sbr_qmf_post_shuffle_c: 112 to 83 Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
* sbrdsp: unroll sbr_autocorrelate_cChristophe Gisquet2013-04-061-0/+25
| | | | | | 1410 cycles to 1148 on Arrandale/Win64. Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
* mips: optimization for float aac decoder (sbr module)Mirjana Vulin2013-02-211-0/+2
| | | | | Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
* x86: call most of the x86 dsp init functions under if (ARCH_X86)Janne Grunau2012-10-081-1/+1
| | | | Rename the called dsp init functions to *_init_x86.
* SBR DSP: unroll sum_squareChristophe GISQUET2012-03-071-4/+9
| | | | | | | | The length is even, so some unrolling can be performed. Timings are for x86: - 32bits: 102c -> 82c - 64bits: 82c -> 69c Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
* SBR DSP x86: implement SSE sbr_sum_square_sseChristophe GISQUET2012-02-231-0/+2
| | | | | | | | | | | | | 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>
* SBR DSP: use intptr_t for the ixh parameter.Christophe GISQUET2012-02-231-1/+1
| | | | Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
* aacsbr: ARM NEON optimised sbrdsp functionsMans Rullgard2012-01-281-0/+4
| | | | | | Overall speedup of HE-AAC decoding 2.3x on Cortex-A8, 1.2x on A9. Signed-off-by: Mans Rullgard <mans@mansr.com>
* aacsbr: move some simdable loops to function pointersMans Rullgard2012-01-281-0/+237
This prepares for assembly optimisations by moving the most time-consuming loops to functions called through pointers in a new context. Signed-off-by: Mans Rullgard <mans@mansr.com>
OpenPOWER on IntegriCloud