mirror of
https://gitlab.freedesktop.org/pipewire/pipewire.git
synced 2026-05-19 21:37:36 -04:00
a7994882b9
Convert matrix_orig and matrix to float arrays and use variable size 2d arrays to access the elements of the matrices. This removes the need for storing pointers to matrix rows.
64 lines
1.7 KiB
C
64 lines
1.7 KiB
C
/* Spa */
|
|
/* SPDX-FileCopyrightText: Copyright © 2018 Wim Taymans */
|
|
/* SPDX-License-Identifier: MIT */
|
|
|
|
#include "channelmix-ops.h"
|
|
|
|
#include <immintrin.h>
|
|
#include <float.h>
|
|
#include <math.h>
|
|
|
|
static inline void clear_avx(float *d, uint32_t n_samples)
|
|
{
|
|
memset(d, 0, n_samples * sizeof(float));
|
|
}
|
|
|
|
static inline void copy_avx(float *d, const float *s, uint32_t n_samples)
|
|
{
|
|
spa_memcpy(d, s, n_samples * sizeof(float));
|
|
}
|
|
|
|
static inline void vol_avx(float *d, const float *s, float vol, uint32_t n_samples)
|
|
{
|
|
uint32_t n, unrolled;
|
|
if (vol == 0.0f) {
|
|
clear_avx(d, n_samples);
|
|
} else if (vol == 1.0f) {
|
|
copy_avx(d, s, n_samples);
|
|
} else {
|
|
__m256 t[4];
|
|
const __m256 v = _mm256_set1_ps(vol);
|
|
|
|
if (SPA_IS_ALIGNED(d, 32) &&
|
|
SPA_IS_ALIGNED(s, 32))
|
|
unrolled = n_samples & ~31;
|
|
else
|
|
unrolled = 0;
|
|
|
|
for(n = 0; n < unrolled; n += 32) {
|
|
t[0] = _mm256_load_ps(&s[n]);
|
|
t[1] = _mm256_load_ps(&s[n+8]);
|
|
t[2] = _mm256_load_ps(&s[n+16]);
|
|
t[3] = _mm256_load_ps(&s[n+24]);
|
|
_mm256_store_ps(&d[n], _mm256_mul_ps(t[0], v));
|
|
_mm256_store_ps(&d[n+8], _mm256_mul_ps(t[1], v));
|
|
_mm256_store_ps(&d[n+16], _mm256_mul_ps(t[2], v));
|
|
_mm256_store_ps(&d[n+24], _mm256_mul_ps(t[3], v));
|
|
}
|
|
for(; n < n_samples; n++) {
|
|
__m128 v = _mm_set1_ps(vol);
|
|
_mm_store_ss(&d[n], _mm_mul_ss(_mm_load_ss(&s[n]), v));
|
|
}
|
|
}
|
|
}
|
|
|
|
void channelmix_copy_avx(struct channelmix *mix, void * SPA_RESTRICT dst[],
|
|
const void * SPA_RESTRICT src[], uint32_t n_samples)
|
|
{
|
|
CHANNELMIX_DEF_MATRIX(matrix, mix, matrix);
|
|
uint32_t i, n_dst = mix->dst_chan;
|
|
float **d = (float **)dst;
|
|
const float **s = (const float **)src;
|
|
for (i = 0; i < n_dst; i++)
|
|
vol_avx(d[i], s[i], matrix[i][i], n_samples);
|
|
}
|