Commit Graph

8973 Commits

Author SHA1 Message Date
Jingning Han
23a82ef5e4 Remove arguments not in use in non-RD decision
This commit removes a few arguments that are not used in the
non-RD mode decision.

Change-Id: I0d878bb5a7dba1b34fe0b64afe55bf11b1a9a2d7
2014-02-27 12:00:24 -08:00
Dmitry Kovalev
5df43a73de Removing redefinition of ROUND_POWER_OF_TWO.
Change-Id: Ic0220f6f3ecdc1c15ba3f93f3ef2b435c0f99b98
2014-02-27 11:57:00 -08:00
Yunqing Wang
e6fd071243 Enable using fast HEX search in non-rd pick mode
Added fast HEX search while doing non-rd partition picking to
speed up the encoder.

Borg test(speed 7) on rtc set showed 1.8% overall PSNR loss.
Encoder speedup was 5% - 15% for different rtc clips.

Change-Id: I9c83026eabc70b69fcc747c90369ec60bfa3ca24
2014-02-27 11:48:07 -08:00
Jingning Han
2c6ddcc606 Cosmetic change mbmi->ref_frame to refs
In handle_inter_mode, the reference frames are set in refs buffer.
One can use refs buffer directly to avoid redundant fetch.

Change-Id: I811d408cae52dcd5e053dd4bfe69550eb6a2ff56
2014-02-27 11:43:03 -08:00
Jingning Han
08053edb80 Assign motion vector invalid value in intra mode
In non-RD mode decision, assign motion vector INVALID_MV when intra
prediction mode is selected.

Change-Id: I8f6ec39b71e755758f0f698074a2c17d934622ea
2014-02-27 11:26:39 -08:00
Jingning Han
bac4113c42 Merge "Run sub-pixel motion search for NEWMV mode test" 2014-02-27 11:04:15 -08:00
Minghai Shang
3a8deeb8b6 Merge "[svc] Add target bitrate settings for each layers." 2014-02-27 10:51:26 -08:00
Minghai Shang
20da0653f4 Merge "Change for adding QP settings for key frames" 2014-02-27 10:50:02 -08:00
Dmitry Kovalev
f81822a568 Merge "Moving common code into vp9_get_entropy_contexts()." 2014-02-27 09:55:46 -08:00
Dmitry Kovalev
129c994f3a Merge "Replacing int_mv with MV in single_motion_search()." 2014-02-27 09:55:18 -08:00
Dmitry Kovalev
516f228d23 Merge "Removing vp9_systemdependent.c." 2014-02-27 09:54:50 -08:00
Deb Mukherjee
01906b53f8 Merge "Refines variance based partitioning search" 2014-02-27 09:52:01 -08:00
Deb Mukherjee
90d13337da Refines variance based partitioning search
Instead of using source variance, this patch uses variance of the
frame difference between the source and the current frame to make
fixed size partition decisions. Also disables adjusting partitioning
if variance based or fixed size partitioning is used.

The latter change improves the speed substantially for speed 6, so
that speed 7 is now less than 3x the speed of speed 6. But speed
6 is 48% better in psnr on the rtc set compared to speed 7.

As compared to speed 5,
speed 6 is -37% in psnr at about 2.5x the speed,
speed 7 is -55% in psnr at about 7x the speed.

Change-Id: If61d80431d3e04ed304ac05832e773cdb2c0a578
2014-02-27 09:37:15 -08:00
Yaowu Xu
962c3bbe11 Merge "vp9_encodemv.h: fix function prototype" 2014-02-27 08:56:32 -08:00
James Zern
cac037635c vp9_encodemv.h: fix function prototype
mismatch was introduced here:
abe21e4 Removing redundant vp9_clear_system_state() call.

Change-Id: I66f945c1ce385dddfdefa9f517f295dbdd01d286
2014-02-26 23:53:24 -08:00
James Zern
af7d99796e Merge "remove ioc problem with rdcost calculation" 2014-02-26 23:45:09 -08:00
James Zern
f0a6a604fc Merge "gen_msvs_vcxproj.sh: Avoid object name collisions." 2014-02-26 23:41:57 -08:00
James Zern
655fa807df Merge changes I03c1ba21,I6d8ac50c,Id0a2d9ef,I3db3acbd,I028bdec9,I26f8724b,I3795d78b,I6bb7ffe8
* changes:
  variance_test: add NEON functions
  subtract_test: add NEON functions
  sixtap_predict_test: add NEON functions
  partial_idct_test: add NEON functions
  fdct8x8_test: add NEON functions
  fdct4x4_test: add NEON functions
  dct32x32_test: add NEON functions
  dct16x16_test: add NEON functions
2014-02-26 23:34:40 -08:00
James Zern
f651bcb296 y4m_video_source: fix memory leak
Begin() will be called twice with 2-pass encodes, invoking
y4m_input_open which allocates memory; close the old instance first.

Change-Id: Id252a21d286ca9ae998bd87599d43aeb8d7d77aa
2014-02-26 23:27:17 -08:00
James Zern
d5e07a8451 variance_test: add NEON functions
note not all functions have NEON implementations:
- variance4x4_neon

Change-Id: I03c1ba21f3b02aa2482d7ca8feedc3ef74b5947f
2014-02-26 19:25:02 -08:00
James Zern
15b57a2bf3 subtract_test: add NEON functions
Change-Id: I6d8ac50ce842749a2a643d8173bbd301cc284236
2014-02-26 19:25:01 -08:00
James Zern
703c79ca17 sixtap_predict_test: add NEON functions
Change-Id: Id0a2d9ef9636c84ee988dabecd5d4dc2b8641944
2014-02-26 19:25:01 -08:00
James Zern
2a66e9f746 partial_idct_test: add NEON functions
note not all functions have NEON implementations:
- vp9_idct32x32_34_add

Change-Id: I3db3acbd5b33839aabd1beadc6e0742ec50b1072
2014-02-26 19:25:01 -08:00
James Zern
c3331105f2 fdct8x8_test: add NEON functions
FwdTrans8x8HT is disabled as the tests currently fail.

note not all functions have NEON implementations:
- fdct8x8/fht8x8

Change-Id: I028bdec9a21eaaee2c5865470ab179aac403540e
2014-02-26 19:25:01 -08:00
James Zern
08c3180732 fdct4x4_test: add NEON functions
Trans4x4HT is disabled as the tests currently fail.

note not all functions have NEON implementations:
- fdct4x4/fht4x4

Change-Id: I26f8724bf2a9ea01d59205a1c57119ed25d043bc
2014-02-26 19:25:00 -08:00
James Zern
a6effda70f dct32x32_test: add NEON functions
note not all functions have NEON implementations:
- vp9_fdct32x32/vp9_fdct32x32_rd

Change-Id: I3795d78bf215e41ab8801c2c37cf633f16931f9d
2014-02-26 19:24:05 -08:00
Alex Converse
104d8ec034 Merge "Don't let GOLDEN_FRAME turn off LAST_FRAME as a reference when GOLDEN_FRAME is disabled." 2014-02-26 18:50:00 -08:00
Dmitry Kovalev
e4118e253a Merge "Adding vpx_sse_to_psnr() function." 2014-02-26 18:38:26 -08:00
Dmitry Kovalev
b83b159e04 Merge "Cleaning up cost_coeffs()." 2014-02-26 18:37:21 -08:00
Dmitry Kovalev
2c594a5275 Removing vp9_systemdependent.c.
Change-Id: I7b9738a7113c0c4687e5d320581ff69d98a8b271
2014-02-26 18:07:23 -08:00
Dmitry Kovalev
a05faf74d4 Replacing int_mv with MV in single_motion_search().
Change-Id: I527c42fa776c772c26c027a47f3dada129f5c33e
2014-02-26 17:07:06 -08:00
Alex Converse
801f136be2 Don't let GOLDEN_FRAME turn off LAST_FRAME as a reference when GOLDEN_FRAME is disabled.
There was a bug in the previous code that if GOLDEN was better than
LAST neither would be used. LAST would get turned off due to superior
GOLDEN quality then all GOLDEN modes would get skipped.

Change-Id: I173f3720451707dab7b2cbbe8b8e6a047089bde7
2014-02-26 16:46:21 -08:00
Ivan Maltz
b2741e9808 Merge "log overall psnr, average psnr, and overall mse" 2014-02-26 16:36:23 -08:00
Dmitry Kovalev
4334c07357 Reusing mem_get_le{16, 32} defined in vpx_ports/mem_opts.h.
Change-Id: If4b5209ac14aaba6f1c1014bc0497baa8eabfaff
2014-02-26 16:32:09 -08:00
Yaowu Xu
2bf9ef4b94 Merge "Reduce the frequency of partition size adjsutment" 2014-02-26 16:23:49 -08:00
Dmitry Kovalev
7d5bffc452 Adding vpx_sse_to_psnr() function.
Removing all copies of identical vp8_mse2psnr/vp9_mse2psnr functions.
Using vpx_sse_to_psnr() instead in all places.

Change-Id: I15beef9834d43d8fc8a8a7a2d1fc5de3d658fed8
2014-02-26 16:21:12 -08:00
Dmitry Kovalev
9349b82bd2 Moving common code into vp9_get_entropy_contexts().
Change-Id: I197670a405f7c2e7796424faa187a61081db8567
2014-02-26 16:20:17 -08:00
Dmitry Kovalev
6232917608 Merge "Cleaning up vp9_encodemb.c file." 2014-02-26 16:03:19 -08:00
Dmitry Kovalev
c313c54835 Cleaning up cost_coeffs().
Change-Id: I02f097fd7c4c787035f03114ddde5b04b2640b39
2014-02-26 14:56:00 -08:00
Marco Paniconi
451ef9051e Merge "Update some comments in vpx_temporal_scalable_patterns." 2014-02-26 14:35:14 -08:00
Marco Paniconi
d32e000ae0 Update some comments in vpx_temporal_scalable_patterns.
Change-Id: I747d26c544cc56feaaf7c75403f2f4c16a496bb7
2014-02-26 14:30:09 -08:00
Ivan Maltz
74e1830b05 log overall psnr, average psnr, and overall mse
Change-Id: I0b4fa1f8104b7fa5780d4e99e5401c14271a992d
2014-02-26 14:18:51 -08:00
Dmitry Kovalev
fc7b7a93aa Replacing int_mv with MV and adding const qualifiers.
Change-Id: I89c37b739493c79044f8c70ae2424aa074eb56c8
2014-02-26 14:13:23 -08:00
Dmitry Kovalev
9774962a24 Cleaning up vp9_encodemb.c file.
Usage of encode_b_args is unnecessary because encode_block_pass1() doesn't
use them. That's why optimize_init_b() call is also not required.

Change-Id: Ib6cfe4916c2ca85749c90bb0adcba6fea592f9ac
2014-02-26 13:51:12 -08:00
Johann
c9a5584e40 Merge "VP8 for ARMv8 by using NEON intrinsics 05" 2014-02-26 13:44:28 -08:00
Dmitry Kovalev
40a65cd1e5 Explicit lossless handling in rd_pick_intra4x4block().
Change-Id: I86cb0be468ade9d891ffd3ed47eceaefac8dd0c4
2014-02-26 13:43:46 -08:00
Minghai Shang
8c196b27b3 [svc] Add target bitrate settings for each layers.
Change-Id: Ia7677fb436667bc4f76db71f65e4784f433f7826
2014-02-26 13:30:50 -08:00
Dmitry Kovalev
28cd972810 Merge "Cleaning up vp9_rdopt.c." 2014-02-26 13:25:10 -08:00
Yaowu Xu
9e5046f521 Merge "Separate super_block_yrd() into two functions" 2014-02-26 13:24:51 -08:00
Johann
801754eead Merge "VP8 for ARMv8 by using NEON intrinsics 04" 2014-02-26 12:28:19 -08:00