Ari Lemmetti
7f873e037c
Updated Makefile to compile picture_x86.asm
2014-07-14 15:30:08 +03:00
Ari Lemmetti
2169f9ab8c
Added AVX asm comments and fixes
...
-Added vzeroupper to satd macro to prevent AVX-SSE transition penalties int picture_x86.asm
-Fixed the order of registers in zero extend macro in picture_x86.asm
-Fixed SATD checkers test pattern in satd_tests.c
2014-07-14 14:43:36 +03:00
Ari Lemmetti
048127c7e3
AVX assembly optimizations improved
2014-07-02 16:57:06 +03:00
Ari Lemmetti
bdef5384ef
Added AVX strategy
2014-06-17 16:52:24 +03:00
Ari Koivula
b3c15b8f94
Merge branch 'owf'
2014-06-16 16:07:41 +03:00
Laurent Fasnacht
91de92134f
Constrain the search not to go under the LCU below if OWF is enabled
2014-06-16 14:27:56 +02:00
Laurent Fasnacht
ef9c2258e9
Fix frame counter and stats
2014-06-16 13:21:52 +02:00
Ari Koivula
153b1ee41f
Merge branch 'intra-sad-strategies'
2014-06-16 12:34:37 +03:00
Laurent Fasnacht
84d34c2655
Fix compilation on non-intel
2014-06-16 11:24:02 +02:00
Ari Koivula
3f00592b96
Separate strategyselector debug prints from _DEBUG.
...
- I only want to see the strategy stuff.
2014-06-16 12:15:19 +03:00
Ari Koivula
1c97a10a6d
Move intra SAD and SATD functions under strategies.
2014-06-16 12:13:41 +03:00
Laurent Fasnacht
4b4702819b
Also print encoding FPS
2014-06-16 11:10:11 +02:00
Laurent Fasnacht
2347574a8e
Fix problems revealed by valgrind
2014-06-16 11:10:09 +02:00
Laurent Fasnacht
28c3f22ba1
Fix possible freeze
2014-06-16 11:03:48 +02:00
Laurent Fasnacht
a96c742ad4
Fix depends for wpp+owf
2014-06-16 11:03:47 +02:00
Laurent Fasnacht
f99e41d41f
Improved CPU time statistics
2014-06-16 11:03:46 +02:00
Laurent Fasnacht
8a33c0a688
Fix recon job for wfrow
2014-06-16 10:55:01 +02:00
Laurent Fasnacht
bf6024734a
Fix statistics with OWF
2014-06-16 10:55:00 +02:00
Laurent Fasnacht
0522a3d8e5
--owf option
2014-06-16 10:55:00 +02:00
Laurent Fasnacht
47d1ded7b0
Dependencies between frames
2014-06-16 10:54:59 +02:00
Laurent Fasnacht
003d3c504c
image_list_copy_contents
2014-06-16 10:54:58 +02:00
Laurent Fasnacht
f4187dd10c
cu_array data structure
2014-06-16 10:54:57 +02:00
Laurent Fasnacht
3be3fa8d6e
Use different processing order depending if we have OWF or not
2014-06-16 10:54:56 +02:00
Laurent Fasnacht
c32943f78b
OWF
2014-06-16 10:54:56 +02:00
Laurent Fasnacht
490dd15f3d
Remove flush between frame
2014-06-16 10:51:33 +02:00
Laurent Fasnacht
fddcbabe28
bitstream writing is now a "normal" job in a thread
2014-06-16 10:51:32 +02:00
Laurent Fasnacht
ff7143cc24
Assign thread_queue_jobs and move image_free to a more suitable place
2014-06-16 10:51:32 +02:00
Ari Koivula
87ca828a63
Correct intra sad function labels.
...
- These haven't been 16 bit for a long time.
2014-06-16 10:45:10 +03:00
Ari Koivula
fcce6ae823
Fix printing of AVX2 capability.
2014-06-14 01:24:19 +03:00
Ari Koivula
a49ba2633a
Add OS and CPU detection for AVX2 and AVX.
2014-06-13 16:57:53 +03:00
Ari Koivula
1de102be61
Move strategies to their own compilation units.
...
- Enforces a little bit more hierarchy. Compilation units are in strategies
and whatever inline includes they have are in a folder with the same name
as the strategy.
2014-06-13 15:30:23 +03:00
Ari Koivula
aa3549a717
Change SLEEP(0) to SLEEP(10) on Windows.
...
- This is a workaround for a performance problem on Windows where main thread
is busy looping.
2014-06-13 12:01:03 +03:00
Laurent Fasnacht
4acadccf89
Only signal the required number of threads
2014-06-13 08:34:59 +02:00
Laurent Fasnacht
70ce7cec20
Remove unneccessary locks by adding threadqueue->queue_running counter
2014-06-13 08:34:58 +02:00
Laurent Fasnacht
7ef34ff5a1
Ability to dump mutex_lock, mutex_unlock and cond_wait timing, if compiled with -D_PTHREAD_DUMP
2014-06-13 08:32:14 +02:00
Laurent Fasnacht
68ad323e84
Tentative fix for race condition
2014-06-12 14:01:33 +02:00
Laurent Fasnacht
b194e19708
Tentative fix for deadlock
2014-06-12 12:57:14 +02:00
Laurent Fasnacht
b765eca153
Remove unneeded encoder_state_blit_pixels
2014-06-12 11:47:46 +02:00
Laurent Fasnacht
da07b8b35d
No-copy works (SAO and deblocking enabled)
2014-06-12 11:47:38 +02:00
Laurent Fasnacht
2cc700fab8
No-copy works with --no-sao (deblocking enabled)
2014-06-12 11:47:31 +02:00
Laurent Fasnacht
6b408b5904
No-copy works with --no-sao --no-deblock
2014-06-12 11:47:30 +02:00
Laurent Fasnacht
0dbfa62698
Replace copy of images made for tiles by sub-images (no copy)
...
- replace width by stride where required in the source code
2014-06-12 11:47:30 +02:00
Laurent Fasnacht
b1347efef5
Add checkpoint in sao_reconstruct
2014-06-12 11:47:29 +02:00
Laurent Fasnacht
ae4dc4eb44
Fix uninitialized sao_info structure members, which was creating false positive when checkpointing SAO
2014-06-12 11:47:29 +02:00
Laurent Fasnacht
f371bdafc3
sao_info checkpoints
2014-06-12 11:47:28 +02:00
Laurent Fasnacht
b7fe81c55c
Checkpoint in pixels_blit, and avoid doing undefined behaviour when source and destination is the same.
...
Seems a reasonnable point to observe when refactoring, since it's called on most image data.
2014-06-12 11:47:28 +02:00
Laurent Fasnacht
da8559fa34
Fix bug in CHECKPOINTS_FINALIZE() when checkpoints are disabled
2014-06-12 11:47:27 +02:00
Laurent Fasnacht
14df6de0d0
Checkpoint on frame checksum
2014-06-12 11:47:00 +02:00
Laurent Fasnacht
22df7cf98b
Use an assert instead of a dumb assignment
2014-06-12 11:47:00 +02:00
Laurent Fasnacht
cf123e317f
Code to checkpoint cu_info and lcu_t
2014-06-12 11:47:00 +02:00