mirrors/pipewire
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/pipewire/pipewire.git synced 2025-11-05 13:30:02 -05:00
Code Issues Projects Releases Packages Wiki Activity Actions

audioconvert: handle more optimizations

Browse source 
Compile an optimized library for the given CPU with the right flags,
then link it with the main library.
This commit is contained in:
Wim Taymans 2019-03-27 17:58:48 +01:00
parent eaffb25cc2
commit c8d3d475bb
12 changed files with 271 additions and 141 deletions
Download patch file Download diff file Expand all files Collapse all files

110
spa/plugins/audioconvert/resample-native-ssse3.h
View file

@ -1,110 +0,0 @@
/* Spa
*
* Copyright © 2019 Wim Taymans
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice (including the next
* paragraph) shall be included in all copies or substantial portions of the
* Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
* DEALINGS IN THE SOFTWARE.
*/
#include <tmmintrin.h>
static void inner_product_ssse3(float *d, const float * SPA_RESTRICT s,
const float * SPA_RESTRICT taps, uint32_t n_taps)
{
__m128 sum = _mm_setzero_ps();
__m128 t0, t1;
uint32_t i;
switch (SPA_PTR_ALIGNMENT(s, 16)) {
case 0:
for (i = 0; i < n_taps; i += 8) {
sum = _mm_add_ps(sum,
_mm_mul_ps(
_mm_load_ps(s + i + 0),
_mm_load_ps(taps + i + 0)));
sum = _mm_add_ps(sum,
_mm_mul_ps(
_mm_load_ps(s + i + 4),
_mm_load_ps(taps + i + 4)));
}
break;
case 4:
t0 = _mm_load_ps(s - 1);
for (i = 0; i < n_taps; i += 8) {
t1 = _mm_load_ps(s + i + 3);
t0 = (__m128)_mm_alignr_epi8((__m128i)t1, (__m128i)t0, 4);
sum = _mm_add_ps(sum,
_mm_mul_ps(t0, _mm_load_ps(taps + i + 0)));
t0 = t1;
t1 = _mm_load_ps(s + i + 7);
t0 = (__m128)_mm_alignr_epi8((__m128i)t1, (__m128i)t0, 4);
sum = _mm_add_ps(sum,
_mm_mul_ps(t0, _mm_load_ps(taps + i + 4)));
t0 = t1;
}
break;
case 8:
t0 = _mm_load_ps(s - 2);
for (i = 0; i < n_taps; i += 8) {
t1 = _mm_load_ps(s + i + 2);
t0 = (__m128)_mm_alignr_epi8((__m128i)t1, (__m128i)t0, 8);
sum = _mm_add_ps(sum,
_mm_mul_ps(t0, _mm_load_ps(taps + i + 0)));
t0 = t1;
t1 = _mm_load_ps(s + i + 6);
t0 = (__m128)_mm_alignr_epi8((__m128i)t1, (__m128i)t0, 8);
sum = _mm_add_ps(sum,
_mm_mul_ps(t0, _mm_load_ps(taps + i + 4)));
t0 = t1;
}
break;
case 12:
t0 = _mm_load_ps(s - 3);
for (i = 0; i < n_taps; i += 8) {
t1 = _mm_load_ps(s + i + 1);
t0 = (__m128)_mm_alignr_epi8((__m128i)t1, (__m128i)t0, 12);
sum = _mm_add_ps(sum,
_mm_mul_ps(t0, _mm_load_ps(taps + i + 0)));
t0 = t1;
t1 = _mm_load_ps(s + i + 5);
t0 = (__m128)_mm_alignr_epi8((__m128i)t1, (__m128i)t0, 12);
sum = _mm_add_ps(sum,
_mm_mul_ps(t0, _mm_load_ps(taps + i + 4)));
t0 = t1;
}
break;
}
sum = _mm_add_ps(sum, _mm_movehl_ps(sum, sum));
sum = _mm_add_ss(sum, _mm_shuffle_ps(sum, sum, 0x55));
_mm_store_ss(d, sum);
}
static void inner_product_ip_ssse3(float *d, const float * SPA_RESTRICT s,
const float * SPA_RESTRICT t0, const float * SPA_RESTRICT t1, float x,
uint32_t n_taps)
{
#if defined (__SSE__)
inner_product_ip_sse(d, s, t0, t1, x, n_taps);
#else
inner_product_ip_c(d, s, t0, t1, x, n_taps);
#endif
}
MAKE_RESAMPLER_FULL(ssse3);
MAKE_RESAMPLER_INTER(ssse3);