]>
granicus.if.org Git - libvpx/log
Scott LaVarnway [Mon, 8 Jun 2015 10:38:13 +0000 (03:38 -0700)]
Update use_prev_frame_mvs flag in decoder.
Added check to see if last frame was all intra. This will
eliminate two checks in find_mv_refs_idx(). Also, do not
update the frame mvs if the current frame is all intra.
This improved performance on material with frequent
intra-only frames.
Change-Id: I44a4042c3670ab0d38439d565062a0e2a1ba9d1e
Parag Salasakar [Fri, 5 Jun 2015 05:32:17 +0000 (05:32 +0000)]
Merge "mips msa vp9 loopfilter 4, 8 optimization"
Parag Salasakar [Fri, 5 Jun 2015 02:22:40 +0000 (07:52 +0530)]
mips msa vp9 loopfilter 4, 8 optimization
average improvement ~3x-4x
Change-Id: I59279293ce4b2a1e99bd10579ac97740e943643f
Marco [Fri, 5 Jun 2015 04:26:05 +0000 (04:26 +0000)]
Merge "Bugfx in setting layer framerate."
Parag Salasakar [Fri, 5 Jun 2015 02:15:24 +0000 (02:15 +0000)]
Merge "mips msa vp9 loopfilter 16 optimization"
James Zern [Fri, 5 Jun 2015 02:15:22 +0000 (02:15 +0000)]
Merge "vp9_reconintra: use AVG[23] consistently"
James Zern [Fri, 5 Jun 2015 01:46:38 +0000 (01:46 +0000)]
Merge "vp9_reconintra_neon_asm/tm4x4: simplify left load"
James Zern [Thu, 4 Jun 2015 22:56:42 +0000 (22:56 +0000)]
Merge "vp9_reconintra: fix d45/d63 discrepancies"
Scott LaVarnway [Thu, 4 Jun 2015 22:20:53 +0000 (22:20 +0000)]
Merge "Remove cm parameter from vp9_decode_block_tokens() part 2"
James Zern [Thu, 4 Jun 2015 20:05:29 +0000 (13:05 -0700)]
vp9_reconintra: fix d45/d63 discrepancies
the final index in rows 2, 3 differ from vp8
Change-Id: I0fcea907b4ab44e266c0f1fd77b290d2236b280a
Marco [Thu, 4 Jun 2015 20:09:44 +0000 (13:09 -0700)]
Bugfx in setting layer framerate.
Index for ts_rate_decimator should be temporal layer (tl) index.
Change-Id: I0320b7f7ae987ef64fdfe7c45099e7978a8fef17
Yaowu Xu [Mon, 1 Jun 2015 16:13:59 +0000 (09:13 -0700)]
Make vp9 the default codec for vpxenc
Change-Id: Ic3b4f3c9a6d8f9b04efb8b5ee080880895063564
Scott LaVarnway [Mon, 1 Jun 2015 15:18:27 +0000 (08:18 -0700)]
Remove cm parameter from vp9_decode_block_tokens() part 2
Change-Id: Iee24b6bb095f748333223e6036fc5c9d9e7e5f1c
Scott LaVarnway [Thu, 4 Jun 2015 13:46:41 +0000 (13:46 +0000)]
Merge "Remove counts param"
Parag Salasakar [Thu, 4 Jun 2015 06:20:41 +0000 (11:50 +0530)]
mips msa vp9 loopfilter 16 optimization
average improvement ~3x-4x
Change-Id: I8ef263da6ebcf8f20aabaefeccf25a84640ba048
Johann Koenig [Thu, 4 Jun 2015 06:16:12 +0000 (06:16 +0000)]
Merge "Make vp9 subpixel match vp8"
Parag Salasakar [Thu, 4 Jun 2015 05:44:23 +0000 (05:44 +0000)]
Merge "mips msa vp9 convolve8 avg hv optimization"
Johann [Tue, 26 May 2015 18:30:25 +0000 (11:30 -0700)]
Make vp9 subpixel match vp8
The only difference between the two was that the vp9 function allowed
for every step in the bilinear filter (16 steps) while vp8 only allowed
for half of those. Since all the call sites in vp9 (<< 1) the input, it
only ever used the same steps as vp8.
This will allow moving the subpel variance to vpx_dsp with the rest of
the variance functions.
Change-Id: I6fa2509350a2dc610c46b3e15bde98a15a084b75
hkuang [Thu, 4 Jun 2015 04:32:09 +0000 (04:32 +0000)]
Merge "Optimize the idct assembly code."
James Zern [Thu, 4 Jun 2015 02:34:37 +0000 (19:34 -0700)]
vp9_reconintra: use AVG[23] consistently
Change-Id: Iab7215f82be0c0c831cd81b6f8091afc3710dd54
Parag Salasakar [Thu, 4 Jun 2015 02:41:01 +0000 (08:11 +0530)]
mips msa vp9 convolve8 avg hv optimization
average improvement ~4x-6x
Change-Id: I7c8b4f2334491be8a859592606e568bc95d019aa
James Zern [Thu, 4 Jun 2015 01:59:09 +0000 (01:59 +0000)]
Merge "vp9_reconintra: simplify d45_predictor"
James Zern [Thu, 4 Jun 2015 01:58:26 +0000 (01:58 +0000)]
Merge changes from topic 'vp9-intra-pred'
* changes:
vp9_reconintra: specialize d135 4x4
vp9_reconintra: specialize d117 4x4
vp9_reconintra: specialize d207 4x4
vp9_reconintra: specialize d153 4x4
vp9_reconintra: specialize d63 4x4
vp9_reconintra: specialize d45 4x4
Marco [Thu, 4 Jun 2015 01:53:43 +0000 (01:53 +0000)]
Merge "Remove ABI check for 1 pass CBR SVC."
James Zern [Thu, 4 Jun 2015 01:51:13 +0000 (18:51 -0700)]
vp9_reconintra_neon_asm/tm4x4: simplify left load
use vld1.8 {d0[]}, [r0] rather than ldrb+vdup; mildly faster
Change-Id: Ia5ffc736bcb0f5497b7d9e55a93bf5a5f5f6928c
Parag Salasakar [Thu, 4 Jun 2015 01:16:18 +0000 (01:16 +0000)]
Merge "mips msa vp9 convolve8 avg horiz optimization"
Marco [Tue, 2 Jun 2015 17:23:03 +0000 (10:23 -0700)]
Remove ABI check for 1 pass CBR SVC.
Remove the ABI check for the controls needed for SVC 1 pass CBR mode.
Bump up the ABI version.
Change-Id: I35b79ee010e14af83c6d1e801d574deaaa2fc7eb
hkuang [Tue, 2 Jun 2015 00:10:33 +0000 (17:10 -0700)]
Optimize the idct assembly code.
Change-Id: Ia0ff859ff1c813dbe100e2f27b1ef78167483f4e
Marco [Wed, 3 Jun 2015 20:52:44 +0000 (13:52 -0700)]
Fix to sample encoder: vpx_temporal_svc_encoder.c
vp8 uses ts_target_bitrate for layer settings.
Change-Id: Ie72477b549051396feebff87a3744fed04366bf4
Johann [Wed, 3 Jun 2015 15:00:51 +0000 (08:00 -0700)]
Remove unused sub pixel mse
Change-Id: I7a5e4e2632c3fa69d2a85a68fa9b418631caf09c
Parag Salasakar [Wed, 3 Jun 2015 06:03:42 +0000 (11:33 +0530)]
mips msa vp9 convolve8 avg horiz optimization
average improvement ~5x-8x
Change-Id: I179a69ec620fbd69979bd128f05d18113618aab4
Parag Salasakar [Wed, 3 Jun 2015 05:58:42 +0000 (05:58 +0000)]
Merge "mips msa vp9 convolve8 avg vert optimization"
James Zern [Wed, 3 Jun 2015 05:49:42 +0000 (05:49 +0000)]
Merge "test-data.sha1: mark test data files as binary"
James Zern [Wed, 3 Jun 2015 05:47:43 +0000 (05:47 +0000)]
Merge "Support building shared libraries on OS/2"
KO Myung-Hun [Mon, 1 Jun 2015 02:35:47 +0000 (11:35 +0900)]
test-data.sha1: mark test data files as binary
Change-Id: Ie3605bf4c4fb16eb21186adbb4577c20a8027344
KO Myung-Hun [Fri, 8 Aug 2014 04:13:02 +0000 (13:13 +0900)]
Support building shared libraries on OS/2
Change-Id: I5750db0504cb69ead52f9f44bf583e693bffbb7e
Parag Salasakar [Wed, 3 Jun 2015 04:25:25 +0000 (09:55 +0530)]
mips msa vp9 convolve8 avg vert optimization
average improvement ~4x-6x
Change-Id: Ia2e6f770da46416ebec31fdcea5cc7878879a9d9
Parag Salasakar [Wed, 3 Jun 2015 04:10:14 +0000 (04:10 +0000)]
Merge "mips msa vp9 idct4x4 and iwht4x4 optimization"
Johann [Wed, 3 Jun 2015 02:03:44 +0000 (02:03 +0000)]
Merge "Unify reference variance functions"
Johann [Mon, 1 Jun 2015 20:03:30 +0000 (13:03 -0700)]
Unify reference variance functions
Use uint32_t for all output and make all functions static
Change-Id: I2c9c6f6310732dc53444607d1c1a268ac1ab83ba
Johann [Tue, 2 Jun 2015 21:44:13 +0000 (21:44 +0000)]
Merge "Disable neon bilinear 4x4"
Scott LaVarnway [Tue, 2 Jun 2015 16:06:00 +0000 (09:06 -0700)]
Remove counts param
Moved to MACROBLOCKD.
Change-Id: Icce765b334f2755f4fe2a4c39fb2ae2d7660d004
Marco [Thu, 21 May 2015 23:15:37 +0000 (16:15 -0700)]
Vidyo patch: Rate control for SVC, 1 pass CBR mode.
-Make Rate control work for SVC 1 pass CBR mode.
-Added temporal layering mode.
-Fixed bug in non-rd variance partition.
-Modified/updated the sample encoders (vp9_spatial_svc_encoder, vpx_temporal_svc_encoder).
-Added datarate unittest(s) for 1 pass CBR SVC.
Change-Id: Ie94b1b68a56ea1267b5087c625e5df04def2ee48
Parag Salasakar [Tue, 2 Jun 2015 06:46:28 +0000 (12:16 +0530)]
mips msa vp9 idct4x4 and iwht4x4 optimization
average improvement ~3x-4x
moved assert to respective files
Change-Id: I6c915059d456a00bdd76fab0dd2eede8b6c6ea58
Parag Salasakar [Tue, 2 Jun 2015 04:59:56 +0000 (10:29 +0530)]
mips msa vp9 updated convolve horiz, vert, hv, copy, avg module
Updated sources according to improved version of common MSA macros.
Enabled respective convolve MSA hooks and tests.
Overall, this is just upgrading the code with styling changes.
Change-Id: If5ad6ef8ea7ca47feed6d2fc9f34f0f0e8b6694d
Parag Salasakar [Tue, 2 Jun 2015 04:48:01 +0000 (04:48 +0000)]
Merge "mips msa vp9 updated idct 8x8, 16x16 and 32x32 module"
Johann [Mon, 1 Jun 2015 23:24:49 +0000 (16:24 -0700)]
Disable neon bilinear 4x4
Clang adds alignment hints when casting up the loads/stores. Although
this should be safe for most paths, it's causing some crashes. Either
the source of the misalignment needs to be determined and adjusted or
the intrinsics need to be rewritten to avoid using the cast to load the
data.
BUG=817,892
Change-Id: Ia3aa824d6a4cd97e14325ff49dc730b6f85ec7e8
Johann Koenig [Mon, 1 Jun 2015 22:36:09 +0000 (22:36 +0000)]
Merge "Adds subsecond frame rates to webm"
James Zern [Mon, 1 Jun 2015 20:03:46 +0000 (20:03 +0000)]
Merge changes from topic 'vp9-intra-pred'
* changes:
vp9_reconintra_neon/tm: improve above_left load
vp9_reconintra_neon: cosmetics: normalize fn params
James Zern [Mon, 1 Jun 2015 20:01:37 +0000 (20:01 +0000)]
Merge "vp9_reconintra_neon_asm/tm: simplify above_left load"
Johann [Mon, 1 Jun 2015 18:56:49 +0000 (18:56 +0000)]
Merge "Cast variance reference output"
paulwilkins [Mon, 1 Jun 2015 18:15:09 +0000 (18:15 +0000)]
Merge "Fast feedback of bits on undershoot."
Johann [Mon, 1 Jun 2015 17:53:46 +0000 (10:53 -0700)]
Cast variance reference output
The larger internal variables are required for the intermediates
but RoundHighBitDepth brings them down to uint32_t/unsigned int.
Fixes type warnings in visual studio.
Change-Id: I48d35284d6cbde330ccdc1f46b6215a645d5eb00
Johnny Klonaris [Mon, 1 Jun 2015 06:29:16 +0000 (23:29 -0700)]
Adds subsecond frame rates to webm
Numerator was being range checked against the
denominator - preventing any frame rate slower
than 1 fps.
I've tested this on a Mac using using ffmpeg and
results are comparable to mp4 and ogg files generated
at the same time.
Not yet tested on Windows.
Johnny Klonaris
google@jawknee.com
Change-Id: Idb358dbc2e7dc000037880ede4a1b0df248a42c8
Parag Salasakar [Mon, 1 Jun 2015 03:49:01 +0000 (09:19 +0530)]
mips msa vp9 updated idct 8x8, 16x16 and 32x32 module
Updated sources according to improved version of common MSA macros.
Enabled idct MSA hooks and tests.
Overall, this is just upgrading the code with styling changes.
Change-Id: I1f488ab2c741f6c622b7a855388a202168082209
James Zern [Sat, 30 May 2015 17:21:15 +0000 (10:21 -0700)]
vp9_reconintra: simplify d45_predictor
only the immediate above right pixel is needed; this removes a
conditional from the inner loop
the final average calculated currently relies on above[] being extended,
it could be reduced to use above[block_size - 2] + 3 * above_right
Change-Id: Ica4f2b8d25eec3ca1d6fa52ef0d4adc228eeea3f
James Zern [Sat, 30 May 2015 03:05:57 +0000 (20:05 -0700)]
vp9_reconintra: specialize d135 4x4
based on webp's RD4()
Change-Id: I64c8f0a1325a8f201eaad39b396fae7a2d06efff
James Zern [Sat, 30 May 2015 03:08:35 +0000 (20:08 -0700)]
vp9_reconintra: specialize d117 4x4
based on webp's VR4()
Change-Id: Ic8c0b8ed65a63772ca0a4321592880a5e8947db5
James Zern [Sat, 30 May 2015 03:11:15 +0000 (20:11 -0700)]
vp9_reconintra: specialize d207 4x4
based on webp's HU4()
Change-Id: I2401ef307cd94e70cc7904f55954af04290c8af9
James Zern [Sat, 30 May 2015 03:12:46 +0000 (20:12 -0700)]
vp9_reconintra: specialize d153 4x4
based on webp's HD4()
Change-Id: Icba1e21ec4b8f5026dc92e49741a68b059c8b9b1
James Zern [Sat, 30 May 2015 03:09:51 +0000 (20:09 -0700)]
vp9_reconintra: specialize d63 4x4
based on webp's VL4()
Change-Id: Ibab962053843eae8752b4e74b6481a53bb034ae9
James Zern [Sat, 30 May 2015 03:01:21 +0000 (20:01 -0700)]
vp9_reconintra: specialize d45 4x4
based on webp's LD4()
Change-Id: I74855d23ce73e1c6988fe08bf7c959b7a69b4abf
Parag Salasakar [Sat, 30 May 2015 02:52:25 +0000 (02:52 +0000)]
Merge "mips msa vp9 updated macros and disable all MSA functions"
James Zern [Sat, 30 May 2015 02:10:27 +0000 (19:10 -0700)]
vp9_reconintra_neon/tm: improve above_left load
use vld1?_dup_u8 over vdup?_n_u8, reduces general register use; mildly
faster
Change-Id: Ie0e4e550849a207b34b378541196b553c9f12011
James Zern [Sat, 30 May 2015 02:01:39 +0000 (19:01 -0700)]
vp9_reconintra_neon: cosmetics: normalize fn params
s/y_stride/stride/
Change-Id: Ie98c3fe241dc240b653849eda356a8862bdd52f4
James Zern [Sat, 30 May 2015 01:56:16 +0000 (18:56 -0700)]
vp9_reconintra_neon_asm/tm: simplify above_left load
use vld1.8 {d0[]}, [r0] rather than ldrb+vdup; mildly faster
Change-Id: I5c24d49a90c2855c94395184774b289da8e9d5a7
James Zern [Sat, 30 May 2015 01:51:35 +0000 (01:51 +0000)]
Merge changes from topic 'vp9-intra-pred'
* changes:
vp9_reconintra_neon: add DC 16x16 predictors
vp9_reconintra_neon: add DC 8x8 predictors
James Zern [Thu, 28 May 2015 03:17:27 +0000 (20:17 -0700)]
vp9_reconintra_neon: add DC 16x16 predictors
85-89% faster over 20M pixels
Change-Id: I9b320ed6b9e67f27df738b84c8b43b65a93c50c2
James Zern [Thu, 28 May 2015 01:59:57 +0000 (18:59 -0700)]
vp9_reconintra_neon: add DC 8x8 predictors
~90% faster over 20M pixels
Change-Id: Iab791510cc57c8332c2f9a5da0ed50702e5f5763
Parag Salasakar [Thu, 28 May 2015 09:59:07 +0000 (15:29 +0530)]
mips msa vp9 updated macros and disable all MSA functions
Done little restructuring/styling changes to the sources like generic macro definitions, their use to reduce code lines, better code alignments etc.
Disabled all MSA hooks and tests
Change-Id: Ic6f2dce0b501f46b80c06c46c0fe2043d557b190
James Zern [Fri, 29 May 2015 06:38:53 +0000 (23:38 -0700)]
variance_neon: add missing include
vpx_ports/mem.h is necessary for MSVC __builtin_prefetch compatibility
macro
Change-Id: I210fad6c6b4545df1874d028b31f42018490b029
Marco [Fri, 29 May 2015 00:21:00 +0000 (17:21 -0700)]
For non-rd pickmode: remove VAR_PARTITION condition.
Keep the logic, transform size based on cyclic refresh and bsize,
(that was conditioned on VAR_PARTITION conditions) the same
for all speeds in non-rd mode (speeds >= 5).
No change to speeds >=6.
Small improvement for speed 5, ~0.5/1.5% gain for avg psnr/ssim.
Change-Id: If9c5657f3d30efd3c7f147166bba7cb69ea55114
Minghai Shang [Thu, 28 May 2015 22:13:53 +0000 (22:13 +0000)]
Merge "[svc] Disable tiles for spatial svc case"
Johann [Thu, 28 May 2015 22:01:53 +0000 (22:01 +0000)]
Merge "Check size restrictions before running test vector"
Scott LaVarnway [Thu, 28 May 2015 19:55:33 +0000 (19:55 +0000)]
Merge "Re-worked header files"
Johann [Thu, 28 May 2015 19:53:50 +0000 (19:53 +0000)]
Merge "Use correct parameters for NEON variance tests"
Johann [Thu, 28 May 2015 19:39:32 +0000 (19:39 +0000)]
Merge "Remove conversion warnings from hbd shifts"
Johann [Thu, 28 May 2015 19:38:56 +0000 (19:38 +0000)]
Merge "Don't #define snprintf in VS 2015 or higher."
Johann [Thu, 28 May 2015 18:28:06 +0000 (11:28 -0700)]
Use correct parameters for NEON variance tests
Change-Id: Ib2949d0a3e9273e7952bbf91956357c1138093f1
hkuang [Thu, 28 May 2015 17:40:58 +0000 (17:40 +0000)]
Merge "Add error handling when running out of free frame buffers."
Johann [Thu, 7 May 2015 23:41:33 +0000 (16:41 -0700)]
Don't #define snprintf in VS 2015 or higher.
In VS 2015 and higher snprintf is supplied and therefore vsnprintf
doesn't need to be defined. This also avoids problems caused by
_snprintf being different from snprintf.
This fixes a build break with VS 2015 and improves security.
Originally submitted via chromium by brucedawson@chromium.org
https://codereview.chromium.org/
1055603003
Additionally break this MSVC-specific tweak to a new file, which will
become the home of all such MSVC-specific things.
This requires adding a dependency on msvc.h to every example which uses
args.c and tools_common.h
Change-Id: I35b5f8e7ea00f6627403aabc9ea79b0412557a99
Johann [Wed, 27 May 2015 21:37:31 +0000 (14:37 -0700)]
Remove conversion warnings from hbd shifts
ROUND_POWER_OF_TWO has some poor side effects when used
with [u]int64_t such as doing the shifting in 32bits.
Change-Id: Ic85a19765cd316fb43657cb21c86f35ceb772773
Johann [Wed, 27 May 2015 22:27:34 +0000 (15:27 -0700)]
Check size restrictions before running test vector
Change-Id: I60ea7724e6ab06fc658f678c1b76d984a43f5a5e
hkuang [Tue, 26 May 2015 18:52:10 +0000 (11:52 -0700)]
Add error handling when running out of free frame buffers.
Change-Id: If28b59b9521204a6e3aecedcf75932d76a752567
Johann [Wed, 27 May 2015 20:34:21 +0000 (20:34 +0000)]
Merge "Only use one 'END' per file"
Minghai Shang [Wed, 27 May 2015 20:24:29 +0000 (20:24 +0000)]
Merge "[decoder] Optimize context buffer re-allocation"
Marco [Wed, 27 May 2015 19:30:35 +0000 (12:30 -0700)]
Non-rd variance partition: Adjust thresholds for 1080p.
Increase the 32x32 split threshold, to allow for more 32x32
at expense of 16x16. Visually looks somewhat better.
Change-Id: Ia1439c3a0dc2d7933468b88bd59266fcd9f03505
Johann [Wed, 27 May 2015 19:26:46 +0000 (19:26 +0000)]
Merge "Correct case in Get4x4SSEFunc"
Marco [Wed, 27 May 2015 19:10:20 +0000 (19:10 +0000)]
Merge "Refactor set_vbp_thresholds."
Johann [Wed, 27 May 2015 19:01:28 +0000 (12:01 -0700)]
Only use one 'END' per file
On visual studio builds the 'END' directive aggressively signals the end
of file.
Change-Id: I28714da32762ef5abcbaeb5a109fb02b80dd13ec
Johann [Wed, 27 May 2015 18:38:43 +0000 (11:38 -0700)]
Correct case in Get4x4SSEFunc
Change-Id: Ie8a7508798fa8e65c579a77cedb8305cee4ddc81
Minghai Shang [Wed, 27 May 2015 17:48:59 +0000 (17:48 +0000)]
Merge "[svc] Make size of empty frame to be 16x16 all the time"
Marco [Thu, 21 May 2015 15:35:11 +0000 (08:35 -0700)]
Refactor set_vbp_thresholds.
Break out the setting of the block variance split thresholds,
since they are locally modified, e.g., based on local/segment qp.
No change in performance.
Change-Id: I0a3238e6dab05140657539fc4bd27ac5ff7a554e
Minghai Shang [Wed, 20 May 2015 23:43:27 +0000 (16:43 -0700)]
[svc] Make size of empty frame to be 16x16 all the time
Change-Id: Ibab09aa0e8c69cf5efea2f0ec035e5da9cc894b0
Johann [Tue, 26 May 2015 23:02:10 +0000 (23:02 +0000)]
Merge "Move variance functions to vpx_dsp"
Johann [Fri, 15 May 2015 18:52:03 +0000 (11:52 -0700)]
Move variance functions to vpx_dsp
subpel functions will be moved in another patch.
Change-Id: Idb2e049bad0b9b32ac42cc7731cd6903de2826ce
Minghai Shang [Tue, 26 May 2015 17:50:42 +0000 (17:50 +0000)]
Merge "[svc] Turn on frame_parallel_decoding_mode"
Jingning Han [Tue, 26 May 2015 16:08:31 +0000 (16:08 +0000)]
Merge "Fix integral projection motion search for frame resize"
James Zern [Sat, 23 May 2015 17:47:51 +0000 (17:47 +0000)]
Merge changes Ie15e301e,Ib070c79b
* changes:
vp9_reconintra_neon: cosmetics: reindent
vp9_reconintra_neon: cosmetics: drop unneeded returns
James Zern [Sat, 23 May 2015 04:04:30 +0000 (21:04 -0700)]
vp9_reconintra_neon: cosmetics: reindent
Change-Id: Ie15e301e8f55cf928f42a03e53a8bb8b66d0e5d5