vpx/vp9/common/vp9_alloccommon.c
hkuang be6aeadaf4 Try again to merge branch 'frame-parallel' into master branch.
In frame parallel decode, libvpx decoder decodes several frames on all
cpus in parallel fashion. If not being flushed, it will only return frame
when all the cpus are busy. If getting flushed, it will return all the
frames in the decoder. Compare with current serial decode mode in which
libvpx decoder is idle between decode calls, libvpx decoder is busy
between decode calls.

Current frame parallel decode will only speed up the decoding for frame
parallel encoded videos. For non frame parallel encoded videos, frame
parallel decode is slower than serial decode due to lack of loopfilter
worker thread.

There are still some known issues that need to be addressed. For example:
decode frame parallel videos with segmentation enabled is not right sometimes.

* frame-parallel:
  Add error handling for frame parallel decode and unit test for that.
  Fix a bug in frame parallel decode and add a unit test for that.
  Add two test vectors to test frame parallel decode.
  Add key frame seeking to webmdec and webm_video_source.
  Implement frame parallel decode for VP9.
  Increase the thread test range to cover 5, 6, 7, 8 threads.
  Fix a bug in adding frame parallel unit test.
  Add VP9 frame-parallel unit test.
  Manually pick "Make the api behavior conform to api spec." from master branch.
  Move vp9_dec_build_inter_predictors_* to decoder folder.
  Add segmentation map array for current and last frame segmentation.
  Include the right header for VP9 worker thread.
  Move vp9_thread.* to common.
  ctrl_get_reference does not need user_priv.
  Seperate the frame buffers from VP9 encoder/decoder structure.
  Revert "Revert "Revert "Revert 3 patches from Hangyu to get Chrome to build:"""
 Conflicts:
       test/codec_factory.h
       test/decode_test_driver.cc
       test/decode_test_driver.h
       test/invalid_file_test.cc
       test/test-data.sha1
       test/test.mk
       test/test_vectors.cc
       vp8/vp8_dx_iface.c
       vp9/common/vp9_alloccommon.c
       vp9/common/vp9_entropymode.c
       vp9/common/vp9_loopfilter_thread.c
       vp9/common/vp9_loopfilter_thread.h
       vp9/common/vp9_mvref_common.c
       vp9/common/vp9_onyxc_int.h
       vp9/common/vp9_reconinter.c
       vp9/decoder/vp9_decodeframe.c
       vp9/decoder/vp9_decodeframe.h
       vp9/decoder/vp9_decodemv.c
       vp9/decoder/vp9_decoder.c
       vp9/decoder/vp9_decoder.h
       vp9/encoder/vp9_encoder.c
       vp9/encoder/vp9_pickmode.c
       vp9/encoder/vp9_rdopt.c
       vp9/vp9_cx_iface.c
       vp9/vp9_dx_iface.c

This reverts commit a18da9760a.

Change-Id: I361442ffec1586d036ea2e0ee97ce4f077585f02
2015-01-30 21:00:13 -08:00

231 lines
6.3 KiB
C

/*
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#include "./vpx_config.h"
#include "vpx_mem/vpx_mem.h"
#include "vp9/common/vp9_blockd.h"
#include "vp9/common/vp9_entropymode.h"
#include "vp9/common/vp9_entropymv.h"
#include "vp9/common/vp9_onyxc_int.h"
#include "vp9/common/vp9_systemdependent.h"
// TODO(hkuang): Don't need to lock the whole pool after implementing atomic
// frame reference count.
void lock_buffer_pool(BufferPool *const pool) {
#if CONFIG_MULTITHREAD
pthread_mutex_lock(&pool->pool_mutex);
#else
(void)pool;
#endif
}
void unlock_buffer_pool(BufferPool *const pool) {
#if CONFIG_MULTITHREAD
pthread_mutex_unlock(&pool->pool_mutex);
#else
(void)pool;
#endif
}
void vp9_set_mb_mi(VP9_COMMON *cm, int width, int height) {
const int aligned_width = ALIGN_POWER_OF_TWO(width, MI_SIZE_LOG2);
const int aligned_height = ALIGN_POWER_OF_TWO(height, MI_SIZE_LOG2);
cm->mi_cols = aligned_width >> MI_SIZE_LOG2;
cm->mi_rows = aligned_height >> MI_SIZE_LOG2;
cm->mi_stride = calc_mi_size(cm->mi_cols);
cm->mb_cols = (cm->mi_cols + 1) >> 1;
cm->mb_rows = (cm->mi_rows + 1) >> 1;
cm->MBs = cm->mb_rows * cm->mb_cols;
}
static int alloc_seg_map(VP9_COMMON *cm, int seg_map_size) {
int i;
for (i = 0; i < NUM_PING_PONG_BUFFERS; ++i) {
cm->seg_map_array[i] = (uint8_t *)vpx_calloc(seg_map_size, 1);
if (cm->seg_map_array[i] == NULL)
return 1;
}
// Init the index.
cm->seg_map_idx = 0;
cm->prev_seg_map_idx = 1;
cm->current_frame_seg_map = cm->seg_map_array[cm->seg_map_idx];
if (!cm->frame_parallel_decode)
cm->last_frame_seg_map = cm->seg_map_array[cm->prev_seg_map_idx];
return 0;
}
static void free_seg_map(VP9_COMMON *cm) {
int i;
for (i = 0; i < NUM_PING_PONG_BUFFERS; ++i) {
vpx_free(cm->seg_map_array[i]);
cm->seg_map_array[i] = NULL;
}
cm->current_frame_seg_map = NULL;
if (!cm->frame_parallel_decode) {
cm->last_frame_seg_map = NULL;
}
}
void vp9_free_ref_frame_buffers(VP9_COMMON *cm) {
BufferPool *const pool = cm->buffer_pool;
int i;
for (i = 0; i < FRAME_BUFFERS; ++i) {
if (pool->frame_bufs[i].ref_count > 0 &&
pool->frame_bufs[i].raw_frame_buffer.data != NULL) {
pool->release_fb_cb(pool->cb_priv, &pool->frame_bufs[i].raw_frame_buffer);
pool->frame_bufs[i].ref_count = 0;
}
vpx_free(pool->frame_bufs[i].mvs);
pool->frame_bufs[i].mvs = NULL;
vp9_free_frame_buffer(&pool->frame_bufs[i].buf);
}
#if CONFIG_VP9_POSTPROC
vp9_free_frame_buffer(&cm->post_proc_buffer);
vp9_free_frame_buffer(&cm->post_proc_buffer_int);
#endif
}
void vp9_free_context_buffers(VP9_COMMON *cm) {
cm->free_mi(cm);
free_seg_map(cm);
vpx_free(cm->above_context);
cm->above_context = NULL;
vpx_free(cm->above_seg_context);
cm->above_seg_context = NULL;
}
int vp9_alloc_context_buffers(VP9_COMMON *cm, int width, int height) {
vp9_free_context_buffers(cm);
vp9_set_mb_mi(cm, width, height);
if (cm->alloc_mi(cm, cm->mi_stride * calc_mi_size(cm->mi_rows)))
goto fail;
// Create the segmentation map structure and set to 0.
free_seg_map(cm);
if (alloc_seg_map(cm, cm->mi_rows * cm->mi_cols))
goto fail;
cm->above_context = (ENTROPY_CONTEXT *)vpx_calloc(
2 * mi_cols_aligned_to_sb(cm->mi_cols) * MAX_MB_PLANE,
sizeof(*cm->above_context));
if (!cm->above_context) goto fail;
cm->above_seg_context = (PARTITION_CONTEXT *)vpx_calloc(
mi_cols_aligned_to_sb(cm->mi_cols), sizeof(*cm->above_seg_context));
if (!cm->above_seg_context) goto fail;
return 0;
fail:
vp9_free_context_buffers(cm);
return 1;
}
static void init_frame_bufs(VP9_COMMON *cm) {
BufferPool *const pool = cm->buffer_pool;
int i;
cm->new_fb_idx = FRAME_BUFFERS - 1;
pool->frame_bufs[cm->new_fb_idx].ref_count = 1;
for (i = 0; i < REF_FRAMES; ++i) {
cm->ref_frame_map[i] = i;
pool->frame_bufs[i].ref_count = 1;
}
}
int vp9_alloc_ref_frame_buffers(VP9_COMMON *cm, int width, int height) {
int i;
const int ss_x = cm->subsampling_x;
const int ss_y = cm->subsampling_y;
vp9_free_ref_frame_buffers(cm);
for (i = 0; i < FRAME_BUFFERS; ++i) {
BufferPool *const pool = cm->buffer_pool;
pool->frame_bufs[i].ref_count = 0;
if (vp9_alloc_frame_buffer(&pool->frame_bufs[i].buf, width, height,
ss_x, ss_y,
#if CONFIG_VP9_HIGHBITDEPTH
cm->use_highbitdepth,
#endif
VP9_ENC_BORDER_IN_PIXELS,
cm->byte_alignment) < 0)
goto fail;
if (pool->frame_bufs[i].mvs == NULL) {
pool->frame_bufs[i].mvs =
(MV_REF *)vpx_calloc(cm->mi_rows * cm->mi_cols,
sizeof(*pool->frame_bufs[i].mvs));
if (pool->frame_bufs[i].mvs == NULL)
goto fail;
pool->frame_bufs[i].mi_rows = cm->mi_rows;
pool->frame_bufs[i].mi_cols = cm->mi_cols;
}
}
init_frame_bufs(cm);
#if CONFIG_VP9_POSTPROC
if (vp9_alloc_frame_buffer(&cm->post_proc_buffer, width, height, ss_x, ss_y,
#if CONFIG_VP9_HIGHBITDEPTH
cm->use_highbitdepth,
#endif
VP9_ENC_BORDER_IN_PIXELS,
cm->byte_alignment) < 0)
goto fail;
#endif
return 0;
fail:
vp9_free_ref_frame_buffers(cm);
return 1;
}
void vp9_remove_common(VP9_COMMON *cm) {
vp9_free_ref_frame_buffers(cm);
vp9_free_context_buffers(cm);
vpx_free(cm->fc);
cm->fc = NULL;
vpx_free(cm->frame_contexts);
cm->frame_contexts = NULL;
}
void vp9_init_context_buffers(VP9_COMMON *cm) {
cm->setup_mi(cm);
if (cm->last_frame_seg_map && !cm->frame_parallel_decode)
vpx_memset(cm->last_frame_seg_map, 0, cm->mi_rows * cm->mi_cols);
}
void vp9_swap_current_and_last_seg_map(VP9_COMMON *cm) {
// Swap indices.
const int tmp = cm->seg_map_idx;
cm->seg_map_idx = cm->prev_seg_map_idx;
cm->prev_seg_map_idx = tmp;
cm->current_frame_seg_map = cm->seg_map_array[cm->seg_map_idx];
cm->last_frame_seg_map = cm->seg_map_array[cm->prev_seg_map_idx];
}