1
0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-23 12:43:46 +02:00
FFmpeg/libavfilter/vf_gblur.c
Ruiling Song 83f9da7768 avfilter/vf_gblur: add x86 SIMD optimizations
The horizontal pass get ~2x performance with the patch
under single thread.

Tested overall performance using the command(avx2 enabled):
./ffmpeg -i 1080p.mp4 -vf gblur -f null /dev/null
./ffmpeg -i 1080p.mp4 -vf gblur=threads=1 -f null /dev/null
For single thread, the fps improves from 43 to 60, about 40%.
For multi-thread, the fps improves from 110 to 130, about 20%.

Signed-off-by: Ruiling Song <ruiling.song@intel.com>
2019-06-12 08:53:11 +08:00

383 lines
13 KiB
C

/*
* Copyright (c) 2011 Pascal Getreuer
* Copyright (c) 2016 Paul B Mahol
*
* Redistribution and use in source and binary forms, with or without modification,
* are permitted provided that the following conditions are met:
*
* * Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* * Redistributions in binary form must reproduce the above
* copyright notice, this list of conditions and the following
* disclaimer in the documentation and/or other materials provided
* with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* HOLDER BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
* EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
* NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
* SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#include "libavutil/imgutils.h"
#include "libavutil/opt.h"
#include "libavutil/pixdesc.h"
#include "avfilter.h"
#include "formats.h"
#include "gblur.h"
#include "internal.h"
#include "video.h"
#define OFFSET(x) offsetof(GBlurContext, x)
#define FLAGS AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
static const AVOption gblur_options[] = {
{ "sigma", "set sigma", OFFSET(sigma), AV_OPT_TYPE_FLOAT, {.dbl=0.5}, 0.0, 1024, FLAGS },
{ "steps", "set number of steps", OFFSET(steps), AV_OPT_TYPE_INT, {.i64=1}, 1, 6, FLAGS },
{ "planes", "set planes to filter", OFFSET(planes), AV_OPT_TYPE_INT, {.i64=0xF}, 0, 0xF, FLAGS },
{ "sigmaV", "set vertical sigma", OFFSET(sigmaV), AV_OPT_TYPE_FLOAT, {.dbl=-1}, -1, 1024, FLAGS },
{ NULL }
};
AVFILTER_DEFINE_CLASS(gblur);
typedef struct ThreadData {
int height;
int width;
} ThreadData;
static void horiz_slice_c(float *buffer, int width, int height, int steps,
float nu, float bscale)
{
int step, x, y;
float *ptr;
for (y = 0; y < height; y++) {
for (step = 0; step < steps; step++) {
ptr = buffer + width * y;
ptr[0] *= bscale;
/* Filter rightwards */
for (x = 1; x < width; x++)
ptr[x] += nu * ptr[x - 1];
ptr[x = width - 1] *= bscale;
/* Filter leftwards */
for (; x > 0; x--)
ptr[x - 1] += nu * ptr[x];
}
}
}
static int filter_horizontally(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
{
GBlurContext *s = ctx->priv;
ThreadData *td = arg;
const int height = td->height;
const int width = td->width;
const int slice_start = (height * jobnr ) / nb_jobs;
const int slice_end = (height * (jobnr+1)) / nb_jobs;
const float boundaryscale = s->boundaryscale;
const int steps = s->steps;
const float nu = s->nu;
float *buffer = s->buffer;
s->horiz_slice(buffer + width * slice_start, width, slice_end - slice_start,
steps, nu, boundaryscale);
emms_c();
return 0;
}
static void do_vertical_columns(float *buffer, int width, int height,
int column_begin, int column_end, int steps,
float nu, float boundaryscale, int column_step)
{
const int numpixels = width * height;
int i, x, k, step;
float *ptr;
for (x = column_begin; x < column_end;) {
for (step = 0; step < steps; step++) {
ptr = buffer + x;
for (k = 0; k < column_step; k++) {
ptr[k] *= boundaryscale;
}
/* Filter downwards */
for (i = width; i < numpixels; i += width) {
for (k = 0; k < column_step; k++) {
ptr[i + k] += nu * ptr[i - width + k];
}
}
i = numpixels - width;
for (k = 0; k < column_step; k++)
ptr[i + k] *= boundaryscale;
/* Filter upwards */
for (; i > 0; i -= width) {
for (k = 0; k < column_step; k++)
ptr[i - width + k] += nu * ptr[i + k];
}
}
x += column_step;
}
}
static int filter_vertically(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
{
GBlurContext *s = ctx->priv;
ThreadData *td = arg;
const int height = td->height;
const int width = td->width;
const int slice_start = (width * jobnr ) / nb_jobs;
const int slice_end = (width * (jobnr+1)) / nb_jobs;
const float boundaryscale = s->boundaryscaleV;
const int steps = s->steps;
const float nu = s->nuV;
float *buffer = s->buffer;
int aligned_end;
aligned_end = slice_start + (((slice_end - slice_start) >> 3) << 3);
/* Filter vertically along columns (process 8 columns in each step) */
do_vertical_columns(buffer, width, height, slice_start, aligned_end,
steps, nu, boundaryscale, 8);
/* Filter un-aligned columns one by one */
do_vertical_columns(buffer, width, height, aligned_end, slice_end,
steps, nu, boundaryscale, 1);
return 0;
}
static int filter_postscale(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
{
GBlurContext *s = ctx->priv;
ThreadData *td = arg;
const int height = td->height;
const int width = td->width;
const int64_t numpixels = width * (int64_t)height;
const unsigned slice_start = (numpixels * jobnr ) / nb_jobs;
const unsigned slice_end = (numpixels * (jobnr+1)) / nb_jobs;
const float postscale = s->postscale * s->postscaleV;
float *buffer = s->buffer;
unsigned i;
for (i = slice_start; i < slice_end; i++)
buffer[i] *= postscale;
return 0;
}
static void gaussianiir2d(AVFilterContext *ctx, int plane)
{
GBlurContext *s = ctx->priv;
const int width = s->planewidth[plane];
const int height = s->planeheight[plane];
const int nb_threads = ff_filter_get_nb_threads(ctx);
ThreadData td;
if (s->sigma <= 0 || s->steps < 0)
return;
td.width = width;
td.height = height;
ctx->internal->execute(ctx, filter_horizontally, &td, NULL, FFMIN(height, nb_threads));
ctx->internal->execute(ctx, filter_vertically, &td, NULL, FFMIN(width, nb_threads));
ctx->internal->execute(ctx, filter_postscale, &td, NULL, FFMIN(width * height, nb_threads));
}
static int query_formats(AVFilterContext *ctx)
{
static const enum AVPixelFormat pix_fmts[] = {
AV_PIX_FMT_YUVA444P, AV_PIX_FMT_YUV444P, AV_PIX_FMT_YUV440P,
AV_PIX_FMT_YUVJ444P, AV_PIX_FMT_YUVJ440P,
AV_PIX_FMT_YUVA422P, AV_PIX_FMT_YUV422P, AV_PIX_FMT_YUVA420P, AV_PIX_FMT_YUV420P,
AV_PIX_FMT_YUVJ422P, AV_PIX_FMT_YUVJ420P,
AV_PIX_FMT_YUVJ411P, AV_PIX_FMT_YUV411P, AV_PIX_FMT_YUV410P,
AV_PIX_FMT_YUV420P9, AV_PIX_FMT_YUV422P9, AV_PIX_FMT_YUV444P9,
AV_PIX_FMT_YUV420P10, AV_PIX_FMT_YUV422P10, AV_PIX_FMT_YUV444P10,
AV_PIX_FMT_YUV420P12, AV_PIX_FMT_YUV422P12, AV_PIX_FMT_YUV444P12, AV_PIX_FMT_YUV440P12,
AV_PIX_FMT_YUV420P14, AV_PIX_FMT_YUV422P14, AV_PIX_FMT_YUV444P14,
AV_PIX_FMT_YUV420P16, AV_PIX_FMT_YUV422P16, AV_PIX_FMT_YUV444P16,
AV_PIX_FMT_YUVA420P9, AV_PIX_FMT_YUVA422P9, AV_PIX_FMT_YUVA444P9,
AV_PIX_FMT_YUVA420P10, AV_PIX_FMT_YUVA422P10, AV_PIX_FMT_YUVA444P10,
AV_PIX_FMT_YUVA420P16, AV_PIX_FMT_YUVA422P16, AV_PIX_FMT_YUVA444P16,
AV_PIX_FMT_GBRP, AV_PIX_FMT_GBRP9, AV_PIX_FMT_GBRP10,
AV_PIX_FMT_GBRP12, AV_PIX_FMT_GBRP14, AV_PIX_FMT_GBRP16,
AV_PIX_FMT_GBRAP, AV_PIX_FMT_GBRAP10, AV_PIX_FMT_GBRAP12, AV_PIX_FMT_GBRAP16,
AV_PIX_FMT_GRAY8, AV_PIX_FMT_GRAY9, AV_PIX_FMT_GRAY10, AV_PIX_FMT_GRAY12, AV_PIX_FMT_GRAY14, AV_PIX_FMT_GRAY16,
AV_PIX_FMT_NONE
};
return ff_set_common_formats(ctx, ff_make_format_list(pix_fmts));
}
void ff_gblur_init(GBlurContext *s)
{
s->horiz_slice = horiz_slice_c;
if (ARCH_X86_64)
ff_gblur_init_x86(s);
}
static int config_input(AVFilterLink *inlink)
{
const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(inlink->format);
GBlurContext *s = inlink->dst->priv;
s->depth = desc->comp[0].depth;
s->planewidth[1] = s->planewidth[2] = AV_CEIL_RSHIFT(inlink->w, desc->log2_chroma_w);
s->planewidth[0] = s->planewidth[3] = inlink->w;
s->planeheight[1] = s->planeheight[2] = AV_CEIL_RSHIFT(inlink->h, desc->log2_chroma_h);
s->planeheight[0] = s->planeheight[3] = inlink->h;
s->nb_planes = av_pix_fmt_count_planes(inlink->format);
s->buffer = av_malloc_array(inlink->w, inlink->h * sizeof(*s->buffer));
if (!s->buffer)
return AVERROR(ENOMEM);
if (s->sigmaV < 0) {
s->sigmaV = s->sigma;
}
ff_gblur_init(s);
return 0;
}
static void set_params(float sigma, int steps, float *postscale, float *boundaryscale, float *nu)
{
double dnu, lambda;
lambda = (sigma * sigma) / (2.0 * steps);
dnu = (1.0 + 2.0 * lambda - sqrt(1.0 + 4.0 * lambda)) / (2.0 * lambda);
*postscale = pow(dnu / lambda, steps);
*boundaryscale = 1.0 / (1.0 - dnu);
*nu = (float)dnu;
}
static int filter_frame(AVFilterLink *inlink, AVFrame *in)
{
AVFilterContext *ctx = inlink->dst;
GBlurContext *s = ctx->priv;
AVFilterLink *outlink = ctx->outputs[0];
AVFrame *out;
int plane;
set_params(s->sigma, s->steps, &s->postscale, &s->boundaryscale, &s->nu);
set_params(s->sigmaV, s->steps, &s->postscaleV, &s->boundaryscaleV, &s->nuV);
if (av_frame_is_writable(in)) {
out = in;
} else {
out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
if (!out) {
av_frame_free(&in);
return AVERROR(ENOMEM);
}
av_frame_copy_props(out, in);
}
for (plane = 0; plane < s->nb_planes; plane++) {
const int height = s->planeheight[plane];
const int width = s->planewidth[plane];
float *bptr = s->buffer;
const uint8_t *src = in->data[plane];
const uint16_t *src16 = (const uint16_t *)in->data[plane];
uint8_t *dst = out->data[plane];
uint16_t *dst16 = (uint16_t *)out->data[plane];
int y, x;
if (!s->sigma || !(s->planes & (1 << plane))) {
if (out != in)
av_image_copy_plane(out->data[plane], out->linesize[plane],
in->data[plane], in->linesize[plane],
width * ((s->depth + 7) / 8), height);
continue;
}
if (s->depth == 8) {
for (y = 0; y < height; y++) {
for (x = 0; x < width; x++) {
bptr[x] = src[x];
}
bptr += width;
src += in->linesize[plane];
}
} else {
for (y = 0; y < height; y++) {
for (x = 0; x < width; x++) {
bptr[x] = src16[x];
}
bptr += width;
src16 += in->linesize[plane] / 2;
}
}
gaussianiir2d(ctx, plane);
bptr = s->buffer;
if (s->depth == 8) {
for (y = 0; y < height; y++) {
for (x = 0; x < width; x++) {
dst[x] = bptr[x];
}
bptr += width;
dst += out->linesize[plane];
}
} else {
for (y = 0; y < height; y++) {
for (x = 0; x < width; x++) {
dst16[x] = bptr[x];
}
bptr += width;
dst16 += out->linesize[plane] / 2;
}
}
}
if (out != in)
av_frame_free(&in);
return ff_filter_frame(outlink, out);
}
static av_cold void uninit(AVFilterContext *ctx)
{
GBlurContext *s = ctx->priv;
av_freep(&s->buffer);
}
static const AVFilterPad gblur_inputs[] = {
{
.name = "default",
.type = AVMEDIA_TYPE_VIDEO,
.config_props = config_input,
.filter_frame = filter_frame,
},
{ NULL }
};
static const AVFilterPad gblur_outputs[] = {
{
.name = "default",
.type = AVMEDIA_TYPE_VIDEO,
},
{ NULL }
};
AVFilter ff_vf_gblur = {
.name = "gblur",
.description = NULL_IF_CONFIG_SMALL("Apply Gaussian Blur filter."),
.priv_size = sizeof(GBlurContext),
.priv_class = &gblur_class,
.uninit = uninit,
.query_formats = query_formats,
.inputs = gblur_inputs,
.outputs = gblur_outputs,
.flags = AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC | AVFILTER_FLAG_SLICE_THREADS,
};