Commit Graph

9 Commits

Author SHA1 Message Date
Janne Grunau
d961a79eb0 sbrdsp: move #if to disable all educational code
Avoids a warning of the unused function 'autocorrelate'.
2014-03-18 13:08:00 +01:00
Christophe Gisquet
4a7af92cc8 sbrdsp: Unroll and use integer operations
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: Diego Biurrun <diego@biurrun.de>
2013-05-03 18:23:30 +02:00
Christophe Gisquet
8394d9a676 sbrdsp: Unroll sbr_autocorrelate_c
1410 cycles to 1148 on Arrandale/Win64

Signed-off-by: Diego Biurrun <diego@biurrun.de>
2013-05-03 18:23:30 +02:00
Janne Grunau
f101eab1be x86: call most of the x86 dsp init functions under if (ARCH_X86)
Rename the called dsp init functions to *_init_x86.
2012-10-08 11:54:05 +02:00
Christophe GISQUET
dabf8dd34a SBR DSP: unroll sum_square
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>
2012-03-07 10:29:52 -08:00
Christophe GISQUET
34454c761f SBR DSP x86: implement SSE sbr_sum_square_sse
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>
2012-02-23 15:50:06 -08:00
Christophe GISQUET
2e74a5abc2 SBR DSP: use intptr_t for the ixh parameter.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2012-02-23 15:48:40 -08:00
Mans Rullgard
be822d77b6 aacsbr: ARM NEON optimised sbrdsp functions
Overall speedup of HE-AAC decoding 2.3x on Cortex-A8, 1.2x on A9.

Signed-off-by: Mans Rullgard <mans@mansr.com>
2012-01-28 14:56:18 +00:00
Mans Rullgard
aac46e088d aacsbr: move some simdable loops to function pointers
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>
2012-01-28 14:56:18 +00:00