Commit Graph

78 Commits

Author SHA1 Message Date
Frank Galligan
e8e152799b Add get release decoder frame buffer functions.
This CL changes libvpx to call a function when a frame buffer
is needed for decode. Libvpx will call a release callback when
no other frames reference the frame buffer. This CL adds a
default implementation of the frame buffer callbacks. Currently
only VP9 is supported. A future CL will add support for
applications to supply their own frame buffer callbacks.

Change-Id: I1405a320118f1cdd95f80c670d52b085a62cb10d
2014-02-10 14:08:11 -08:00
Dmitry Kovalev
88340b173b Merge "Combining fb_idx_ref_cnt[] and yv12_fb[] arrays." 2014-01-31 15:55:04 -08:00
Yunqing Wang
903801f1ef vp9 decoder: row-based multi-threaded loopfilter
Implemented parallel loopfiltering, which uses existing tile-
decoding threads. Each thread works on one row, and when that row
is loopfiltered, it moves to next unattended row. To ensure the
correct filtering order, threads are synchronized and one
superblock is filtered only if the superblocks it depends on are
filtered already.

To reduce synchronization overhead and speed up the decoder, we use
nsync > 1 for high resolution.

Performance tests:
1. on desktop:
8-tile 4k video using 8 threads, speedup: 70% - 80%
4-tile HD video using 4 threads, speedup: ~35%
2. on mobile device(Nexus 7):
4-tile 1080p video using 4 threads, speedup: 18% - 25%
4-tile 1080p video using 2 threads, speedup: 10% - 15%

Change-Id: If54b4a11960dd706c22d5ad145ad94156031f36a
2014-01-31 14:44:53 -08:00
Adrian Grange
64e2597465 Disable update of last_show_frame for existing frame
When showing a previously decoded frame, i.e. when
show_existing_frame=1, the update of the
last_show_frame flag must be disabled.

This is to ensure that the last_show_frame flag
reflects the state of the flag for the immediately
previously decoded frame rather then the value that
was forced to ensure that a previously decoded frame
would be displayed.

This patch also adds a test vector to verify that the
display_existing_frame flag works as expected. Code
for generating the test vector can be found in this
patch:
https://gerrit.chromium.org/gerrit/#/c/68581/

(Bug originally reported by Alexander Voronov
<ru.xalba@gmail.com>).

Change-Id: I731d288fba02088959f7fcc87707137fffc6acf5
2014-01-31 08:55:01 -08:00
Dmitry Kovalev
6332063475 Combining fb_idx_ref_cnt[] and yv12_fb[] arrays.
Adding new RefCntBuffer struct which contains reference counter and image
buffer.

Change-Id: I71c1f532faa13442c32c43fc03ec45b6f88fb844
2014-01-29 12:48:01 -08:00
Dmitry Kovalev
ba41e9d459 Adding RefBuffer struct.
Adding RefBuffer to simplify reference buffer management. The struct has a
pointer to image data and scale factors relative to the current frame.

Change-Id: If38eb1491ff687cc11428aee339f3e052e2c5d9e
2014-01-03 15:21:55 -08:00
Yaowu Xu
8458c8c450 Merge "Fix show existing frame" 2014-01-02 09:27:28 -08:00
hkuang
ff2c96be1f Implenment on demand border extension. In place extend
the border now. Next commit will totally remove the border.

Change-Id: Ic1e1ca9cc34f81c688715b3948689b47df63a151
2013-12-09 16:44:08 -08:00
Dmitry Kovalev
0d4b8d7e43 Renaming constants.
NUM_YV12_BUFFERS        => FRAME_BUFFERS
ALLOWED_REFS_PER_FRAME  => REFS_PER_FRAME
NUM_REF_FRAMES_LOG2     => REF_FRAMES_LOG2
NUM_REF_FRAMES          => REF_FRAMES
NUM_FRAME_CONTEXTS_LOG2 => FRAME_CONTEXTS_LOG2
NUM_FRAME_CONTEXTS      => FRAME_CONTEXTS

Change-Id: I4e1ada08f25d8fa30fdf03aebe1b1c9df0f87e63
2013-12-05 16:23:09 -08:00
Tero Rintaluoma
047b0b01bb Fix show existing frame
- Disable mode info update in case where current frame is coded
  as "show existing frame".
- Should fix issue 676.

Change-Id: Ibee681850eb307f982da6528d3e31cb94f881c08
2013-12-05 12:10:10 +02:00
Frank Galligan
9ed616a56c Fix the initial references to frame buffers.
The old code would start in a mixed state, where all the reference
frames were pointing to frame buffer 0, but the reference counts
were 0. This is why we needed special code for the first frame.

Change-Id: I734961012917654ff8c0c8b317aac00ab75ded1a
2013-12-04 16:53:18 -08:00
Dmitry Kovalev
f00d157c12 Moving eob array to the encoder.
In the decoder we don't need to save eobs, we can pass eob as an argument.
That's why removing eob arrays from VP9Decompressor and TileWorkerData,
and moving eob pointer from macroblockd_plane to macroblock_plane.

Change-Id: I8eb919acc837acfb3abdd8319af63d1bbca8217a
2013-12-03 17:59:32 -08:00
Dmitry Kovalev
d3a2e55af4 Removing qcoeff buffers from the decoder.
We only need qcoeff buffers in the encoder. Reducing TileWorkerData struct
and VP9Decompressor struct sizes by 24K.

Change-Id: Id148868461f7ffa3d3dd634b371503ae9c57e207
2013-11-26 18:52:10 -08:00
Yaowu Xu
49cbe4580d Renamed two files
from vp9_decodframe.{c,h} to vp9_decodeframe.{c,h}

Change-Id: I21ac4b14fc90246e3f16bd90c52c12d126d791f8
2013-11-15 12:48:43 -08:00
Jingning Han
d8b4c79270 Decouple macroblockd_plane buffer usage
Make the macroblockd_plane contain dynamic buffer pointers instead
static pointers to the memory space allocated therein. The decoder
uses the buffer allocated in pbi, while encoder will use a dual
buffer approach for rate-distortion optimization search.

Change-Id: Ie6f24be2dcda35df7c15b4014e5ccf236fb3f76c
2013-11-11 15:26:10 -08:00
James Zern
fb484524bd vp9: add multi-threaded tile decoder
tiles are decoded in parallel within a single frame

Change-Id: I7aca87cb1c239b74eceef72bdc9f672faebac373
2013-10-30 01:00:20 +01:00
James Zern
3ffa41aae3 Merge changes If9b16f7d,I75aab21c,I9cbb768c,If5cea3d3,I96940657,I025595d8,Ie0bc3935,I3ebb172d
* changes:
  vp9: remove partition+entropy contexts from common
  vp9: add above/left_context to MACROBLOCKD
  vp9: add above/left_seg_context to MACROBLOCKD
  vp9: add above/left_context to encoder
  vp9: add above/left_seg_context to encoder
  vp9: pass entropy context directly to set_skip_context
  vp9: pass context directly to partition functions
  vp9/decode: add alloc_tile_storage()
2013-10-28 12:45:11 -07:00
James Zern
e571d3badc vp9: add above/left_context to MACROBLOCKD
Change-Id: I75aab21c1692cbad717564cbb436578fddbc348d
2013-10-28 11:34:18 +01:00
James Zern
d9a317c8b2 vp9: add above/left_seg_context to MACROBLOCKD
Change-Id: I9cbb768c5f857a096cf6c29d6755d0e5e6728435
2013-10-28 11:32:16 +01:00
James Zern
8f177bb0b6 vp9 decode: defer loop filter allocation
wait until do_loopfilter_inline is true before committing the resources

Change-Id: I01661bd40599b47362bb3fb534668471f2a9d8d7
2013-10-26 11:57:44 +02:00
Dmitry Kovalev
237ce8724a Adding get_frame_new_buffer() function to replace duplicated code.
Change-Id: I6e0e19231a48364c1de7dfab730b121ab227f111
2013-10-24 12:20:35 -07:00
Dmitry Kovalev
dfc7945d1e Adding get_frame_ref_buffer() function + cleanup.
Change-Id: Ib9ead216fc54b2df6f6f1fe82d2ea137197beebd
2013-10-24 11:05:35 -07:00
James Zern
f8797a5aac vp9/dec: use loop filter worker for all inline cases
- uses vp9_worker_execute() in non-threaded case
+ removes hook reassignment

Change-Id: I305fd55f2be5cd92367373c929cdb9004066f90f
2013-10-21 14:28:40 +02:00
Guillaume Martres
5b984b36ca Use a separate MODE_INFO stream for each tile column
This should make parallel tiles decoding easier to implement.

Change-Id: I6226456dd11f275fa991e4a7a930549da6675915
2013-10-16 16:24:48 -07:00
Guillaume Martres
acf0d56f0b Get rid of "this_mi", use "mi_8x8[0]" everywhere instead
The only case where they were intentionally pointing to different
structures was in mbgraph, and this didn't have the expected behavior
because both of these pointers are used interchangeably through the code

Change-Id: I979251782f90885fe962305bcc845bc05907f80c
2013-10-16 16:24:03 -07:00
Guillaume Martres
42bcb4a7ad Merge "Prevent accidental changes to the previous frame mode_infos" 2013-10-16 16:07:05 -07:00
Guillaume Martres
b364176c08 Prevent accidental changes to the previous frame mode_infos
This is needed to fix mbgraph but shouldn't affect anything else

Change-Id: I2f515052f62e348cd3794b7ff0c139802225ea95
2013-10-10 12:18:12 -07:00
James Zern
557862d152 vp9_receive_compressed_data: remove unnecessary indent
+ useless comment

Change-Id: Ied29a4cc8c506b216968ce67af630bae542aca12
2013-10-05 12:10:38 +02:00
Dmitry Kovalev
1a9d4fedf3 Merge "Using size_t for memory buffer size." 2013-09-30 11:10:08 -07:00
Jim Bankoski
cfbc246d57 fixed cpplint issues in vp9_onyxd_if.c
Change-Id: Ia67e9ed2d5ea79f3dbf1d58f9a187cb18ecd0995
2013-09-29 11:03:53 -07:00
Dmitry Kovalev
c7b7b1da86 Using size_t for memory buffer size.
Change-Id: Ibf1642525731c66c99fa25f95c7b5834ae88c688
2013-09-24 16:38:30 -07:00
Scott LaVarnway
ac6093d179 New mode_info_context storage -- undo revert
mode_info_context was stored as a grid of MODE_INFO structs.
The grid now constists of pointers to MODE_INFO structs.  The
MODE_INFO structs are now stored as a stream (decoder only),
eliminating unnecessary copies and is a little more cache
friendly.

Change-Id: I031d376284c6eb98a38ad5595b797f048a6cfc0d
2013-09-11 13:45:44 -04:00
James Zern
54a03e20dd Revert "New mode_info_context storage"
This reverts commit dae17734ec

Encode crashes, leaks and increases integer overflow errors.

Change-Id: I595aa2649bb8d0b6552ff91652837a74c103fda2
2013-09-09 13:37:01 -07:00
Scott LaVarnway
dae17734ec New mode_info_context storage
mode_info_context was stored as a grid of MODE_INFO structs.
The grid now constists of a pointer to a MODE_INFO struct and
a "in the image" flag.  The MODE_INFO structs are now stored
as a stream, eliminating unnecessary copies and is a little
more cache friendly.

For the test clips used, the decoder performance improved
by ~4.3% (1080p) and ~9.7% (720p).

Patch Set 2: Re-encoded clips with latest. Now ~1.7% (1080p)
and 5.9% (720p).

Change-Id: I846f29e88610fce2523ca697a9a9ef2a182e9256
2013-09-06 12:33:34 -04:00
Jim Bankoski
79401542f7 make vp9 postproc a config option
Vp9 postproc is disabled for now as its not been shown to help and
may be merged with vp8.

Change-Id: I25620d6cd34c6e10331b18c7b5ef7482e39c6057
2013-09-04 10:02:08 -07:00
Adrian Grange
3e340880a8 Merge "Added resizing & initialization of last frame segment map" 2013-08-16 09:07:36 -07:00
Adrian Grange
d5bec522da Added resizing & initialization of last frame segment map
When the frame size changes the last frame segment map must
be resized to match and initialized to 0.

Change-Id: Idc10de109f55dbe9af3a6caae355a2974712243d
2013-08-15 15:35:21 -07:00
Dmitry Kovalev
67fe9d17cb Removing redundant code and function arguments.
Change-Id: Ia5cdda0f755befcd1e64397452c42cb7031ca574
2013-08-09 17:24:40 -07:00
Dmitry Kovalev
816d6c989c Moving loopfilter struct to VP9_COMMON.
Loop filter configuration doesn't belong to macroblock, so moving it from
MACROBLOCKD to VP9_COMMON. Also moving the declaration of loopfilter struct
from vp9_blockd.h to vp9_loopfilter.h.

Change-Id: I4b3e34be9623b47cda35f9b1f9951f8c5b1d5d28
2013-08-09 14:41:51 -07:00
Yaowu Xu
6ec2b85bad Added lpf level picking using partial frame
Change-Id: I599ab1bd22b5f3f10d5962c609952abdef8ff67a
2013-08-09 07:37:08 -07:00
Dmitry Kovalev
6a8ec3eac2 General code cleanup.
Removing redundant parenthesis and curly braces. Combining declarations
with initializations. Adding useful intermediate variables instead of
recalculating expressions every time.

Change-Id: I00106f404afd60bfc189905b0fded881684f941a
2013-08-08 21:12:34 -07:00
James Zern
a0ffa2794b vp9/decoder: threaded row-based loop filter
Currently the only threaded option for vp9 decode. Enabled when the
decoder config thread count is > 1.

Change-Id: I082959abac9e31aa4a38ed9fd68b94680e57f4df
2013-08-05 13:22:04 -07:00
Dmitry Kovalev
7131cb0e3d General cleanups.
Removing unused constants, macros, and function declarations. Using
ROUND_POWER_OF_TWO macro, vp9_zero, vp9_copy where possible. Moving
#include from *.h to *.c. Merging for loops for motion vectors.

Change-Id: Ic3bf841764a2bb177128bb3a6d7aa8f68229cd13
2013-07-25 14:13:48 -07:00
James Zern
38a4412e1b vp9: apply loopfilter inline if possible
excludes tiled content currently

Change-Id: I44155253e8d6771e5e039d663be5f21cc9d0355d
2013-07-22 15:52:10 -07:00
Dmitry Kovalev
ee1771ebaa Moving all loop filter related variables into new struct.
Adding loopfilter struct with fields from MACROBLOCKD and VP9Common.
Eventually it will be moved to vp9_loopfilter.h for better code structure.

Change-Id: Iaf5fb71c33719cdfa1b991f671caf071be9ea035
2013-07-19 16:19:10 -07:00
Dmitry Kovalev
3de20f3ea9 Removing unused members of VP9Decompressor: mbc, prob_skip_false.
Change-Id: Id5480a4fd56c184ad046c2192b30d190debb3de0
2013-07-18 11:29:34 -07:00
James Zern
3a7c2665d0 Merge "yv12config: remove YUV_TYPE" 2013-07-16 12:16:04 -07:00
Yaowu Xu
5b915ebd92 Change to extend full border only when needed
This is a short term optimization till we work out a decoder
implementation requiring no frame border extension.

Change-Id: I02d15bfde4d926b50a4e58b393d8c4062d1be70f
2013-07-15 20:52:13 -07:00
James Zern
4fc6c88e9c yv12config: remove YUV_TYPE
this was never fleshed out in the context of VP8, for which it was
added. for VP9 it has no meaning.

Change-Id: Iba2ecc026d9e947067b96690245d337e51e26eff
2013-07-12 15:25:48 -07:00
Ronald S. Bultje
2afc3422c6 Remove unused and outdated debug code.
Change-Id: I0e789bdeaed60f920f7a470e56a8d4ea374233fc
2013-05-28 19:15:57 -07:00