system/pipewire
system/pipewire
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/pipewire/pipewire synced 2024-10-01 13:44:40 +00:00
Code Issues Packages Projects Releases Wiki Activity

filter-chain: improve convolver

Browse source 
Make the input buffer a little larger and leave the top blockSize
samples zero. That way we can fill up the lower part, leave the upper
part zero padded and feed this to the fft directly. Also only clear
the lower part when we can't fill it completely.

This removes some memcpy and memset operations.
This commit is contained in:
Wim Taymans 2021-08-26 17:16:36 +02:00
parent 1924c2c298
commit 73854e9ec9
1 changed files with 7 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

14
src/modules/module-filter-chain/convolver.c
View file

@ -189,7 +189,7 @@ static struct convolver1 *convolver1_new(int block, const float *ir, int irlen)
fft_cpx_init(&conv->pre_mult, conv->fftComplexSize);
fft_cpx_init(&conv->conv, conv->fftComplexSize);
conv->overlap = fft_alloc(sizeof(float) * conv->blockSize);
conv->inputBuffer = fft_alloc(sizeof(float) * conv->blockSize);
conv->inputBuffer = fft_alloc(sizeof(float) * conv->segSize);
conv->inputBufferFill = 0;
conv->current = 0;
@ -229,11 +229,11 @@ static int convolver1_run(struct convolver1 *conv, const float *input, float *ou
const int inputBufferPos = conv->inputBufferFill;
memcpy(conv->inputBuffer + inputBufferPos, input + processed, processing * sizeof(float));
if (inputBufferPos == 0 && processing < conv->blockSize)
memset(conv->inputBuffer + processing, 0,
sizeof(float) * (conv->blockSize - processing));
memcpy(conv->fft_buffer, conv->inputBuffer, conv->blockSize * sizeof(float));
memset(conv->fft_buffer + conv->blockSize, 0, (conv->segSize - conv->blockSize) * sizeof(float));
fft_run(conv->fft, conv->fft_buffer, &conv->segments[conv->current]);
fft_run(conv->fft, conv->inputBuffer, &conv->segments[conv->current]);
if (conv->inputBufferFill == 0) {
fft_cpx_clear(&conv->pre_mult, conv->fftComplexSize);
@ -255,11 +255,11 @@ static int convolver1_run(struct convolver1 *conv, const float *input, float *ou
ifft_run(conv->ifft, &conv->conv, conv->fft_buffer);
fft_sum(output + processed, conv->fft_buffer + inputBufferPos, conv->overlap + inputBufferPos, processing);
fft_sum(output + processed, conv->fft_buffer + inputBufferPos,
conv->overlap + inputBufferPos, processing);
conv->inputBufferFill += processing;
if (conv->inputBufferFill == conv->blockSize) {
memset(conv->inputBuffer, 0, sizeof(float) * conv->blockSize);
conv->inputBufferFill = 0;
memcpy(conv->overlap, conv->fft_buffer + conv->blockSize, conv->blockSize * sizeof(float));