Граф коммитов

12505 Коммитов

Автор SHA1 Сообщение Дата
Johann 37ea1e1218 Merge "VP9 common for ARMv8 by using NEON intrinsics 05" 2014-12-15 14:48:53 -08:00
Jingning Han 5c93dca3d3 Merge "Simplify rate-distortion modeling function" 2014-12-15 14:37:19 -08:00
Jingning Han c2c7596fc7 Initialize best_tx_size with invalid value
If vp9_pick_inter_mode works properly, it should at least check
one coding mode and hence get best_tx_size assigned a valid value.
There is no need to initialize best_tx_size with a legitimate
value before starting the mode search.

Change-Id: Ic0496cd89672ea9c2c512a9bd1da952190af9cba
2014-12-15 12:58:34 -08:00
Jingning Han 83e2c62aba Use right shift to replace division in vp9_pick_inter_mode
Make the variable reduction_fac log2 based and explicitly use
right shift when computing intra_cost_penalty.

Change-Id: I208f1fb879a02debb3b3fc64f9fd06260dcf1c86
2014-12-15 12:48:07 -08:00
Frank Galligan c4f7079ad4 Revert "Revert "Add support for setting byte alignment.""
This reverts commit 91471d6aad.

Fixes the compile issues if post_proc is enabled.

Change-Id: Ib40a15ce2c194f9b5adfa65a17ab01ddf60f5a59
2014-12-15 12:20:37 -08:00
James Yu 4f856cd7fa VP9 common for ARMv8 by using NEON intrinsics 06
Add vp9_iht8x8_add_neon.c
- vp9_iht8x8_64_add_neon

The assembly did not previously implement tx_type 0
BUG=716

Change-Id: Icfc99dd24f3d59047f9184a7d0c761ba7e3de934
Signed-off-by: James Yu <james.yu@linaro.org>
2014-12-15 12:18:06 -08:00
James Yu 6b71013277 VP9 common for ARMv8 by using NEON intrinsics 05
Add vp9_iht4x4_add_neon.c
- vp9_iht4x4_16_add_neon

The assembly did not previously implement tx_type 0
BUG=715

Change-Id: I60034d1568de034edba45c5cdd13f3d87dbc73b6
Signed-off-by: James Yu <james.yu@linaro.org>
2014-12-15 12:16:19 -08:00
James Zern 8d558f2ca5 Merge "vp9/MACROBLOCKD: reorder struct members" 2014-12-15 11:54:51 -08:00
Jingning Han eefe869291 Simplify rate-distortion modeling function
Use left shift to replace one multiplication. The computation
outcome remains identical.

Change-Id: I1e1737af0a245de0d2a2bde10f0c171477199fc1
2014-12-15 11:51:16 -08:00
Paul Wilkins 91471d6aad Revert "Add support for setting byte alignment."
Fails to compile. Bad calls to vp9_alloc_frame_buffer
and vp9_realloc_frame_buffer in postproc.c

This reverts commit 399823b6f5.

Change-Id: I29f0e173f8e185d3a303cfdb17813e1eccb51e3a
2014-12-15 11:54:13 +00:00
hkuang d4426d5876 Merge "Fix a bug that break the vp8 fragment decoder." 2014-12-13 09:56:55 -08:00
James Zern d356dce759 iosbuild: add success/failure output
Change-Id: I84492f68752321f0266141666e2672ed2da5f509
2014-12-12 19:45:15 -08:00
James Zern c58c579ec4 vp9/MACROBLOCKD: reorder struct members
improves locality of reference

Change-Id: I0639b98bf38879f918173b3a1b25dd93090e88b4
2014-12-12 18:01:24 -08:00
hkuang 9353607ac8 Fix a bug that break the vp8 fragment decoder.
(issue #882).

Change-Id: I2ca7f96d390c4eaec0473c50cb01b903d0bd3ee6
2014-12-12 16:44:14 -08:00
James Zern 089086bc25 Merge "Optimize bit_read_buffer." 2014-12-12 16:29:42 -08:00
Tom Finegan b3345f462f Merge "vpxdec: Rename the libyuv scale wrapper." 2014-12-12 16:14:57 -08:00
Tom Finegan 53a0ed84ac Merge "iosbuild.sh: Add targets argument." 2014-12-12 16:14:12 -08:00
Frank Galligan 9c2601eb68 Merge "Add support for setting byte alignment." 2014-12-12 15:47:11 -08:00
hkuang 3cecce916b Optimize bit_read_buffer.
Change-Id: Iee43c34909deec9787b29c1c33672213b9f049df
2014-12-12 14:38:12 -08:00
James Zern 89ee8923a8 Merge "Remove redundant loads on 1d16_v8 filter." 2014-12-12 14:32:52 -08:00
James Zern f82d7fd854 Merge "Remove redundant loads on 1d8_v8 filter." 2014-12-12 14:32:26 -08:00
James Zern 4d40a046da Merge "vp9: move encoder-only member from common" 2014-12-12 14:28:55 -08:00
James Zern 0c0a57b94d Merge "don't set INLINE to 'always_inline'" 2014-12-12 14:28:20 -08:00
James Zern 2bf4b4852f Merge changes Id6421838,I37499329
* changes:
  vp9: make postproc members depend on CONFIG_VP9_POSTPROC
  vp9_postproc: remove redundant CONFIG_* checks
2014-12-12 14:27:56 -08:00
Marco 7f59cff53d Merge "Allow for 4x4 prediction blocks for key frame, speed 6." 2014-12-12 14:27:31 -08:00
James Zern 5ccff43292 Merge "vp9_loopfilter_mmx: remove some unused tables" 2014-12-12 14:25:53 -08:00
James Zern 3aecec084b Merge "x86_abi_support: set LIBVPX_RAND w/vp9-postproc" 2014-12-12 14:25:30 -08:00
Tom Finegan bc74a2f33b iosbuild.sh: Add targets argument.
Allows override of default target list. Also added missing usage info
for --extra-configure-args, and removed last vestiges of armv6 support.

Change-Id: Ic0f14fffa0cbaea1bed371d38ff65e035bbe3273
2014-12-12 13:53:58 -08:00
Frank Galligan 399823b6f5 Add support for setting byte alignment.
Add support for setting byte alignment on the Y, U, and V plane of the
reference buffers. The byte alignment must be a power of 2, from 32 to
1024. A value of 0 sets legacy alignment.

Change-Id: I7c1399622f7aa68e123646369216b32047dda73d
2014-12-12 13:34:36 -08:00
James Zern 6d1a63a02a Merge "Remove unnecessary dqcoeff memset." 2014-12-12 12:16:32 -08:00
Tom Finegan 8ef2ce8664 vpxdec: Rename the libyuv scale wrapper.
The other name was misleading: We do not export scaling support from
libvpx via vpx_im{g,age}*.

Change-Id: I8acb4ea0301f08c9bab557a4063ea35d147b4631
2014-12-12 11:58:53 -08:00
Frank Galligan 6a24dbd71f Remove redundant loads on 1d16_v8 filter.
This CL showed about a 3% gain in performance on some systems.

Change-Id: Id27e7e0b8e69068aa364e67859436da852669250
2014-12-12 11:48:47 -08:00
Frank Galligan 44ee777905 Remove redundant loads on 1d8_v8 filter.
This CL showed a modest gain in performance on some systems.

Change-Id: Iad636a89a1a9804ab7a0dea302bf2c6a4d1653a4
2014-12-12 11:34:24 -08:00
James Zern 5fe9a5c649 don't set INLINE to 'always_inline'
INLINE is used quite widely in vp9, this change improves performance
1-2% on most modern platforms.

Change-Id: I8a9974aab89fa588ea4923cc7eaf6199e344a528
2014-12-12 11:18:00 -08:00
James Zern 72ece1308b vp9: move encoder-only member from common
allow_comp_inter_inter VP9_COMMON -> VP9_COMP

Change-Id: I6d9dc25d1cdd7e2ab62f5be69cd9fa883d21dbb6
2014-12-12 11:17:44 -08:00
James Zern ef06de33fe vp9: make postproc members depend on CONFIG_VP9_POSTPROC
Change-Id: Id64218386968cee3132269e4a0572650f20fd980
2014-12-12 11:17:17 -08:00
James Zern 890f7bedf3 vp9_postproc: remove redundant CONFIG_* checks
the entire module is wrapped in CONFIG_VP9_POSTPROC which is forcibly
enabled with CONFIG_INTERNAL_STATS

+ a similar change in vp9_alloccommon.c

Change-Id: I374993297a9fba5bef2f0b71f984eba42f0995a3
2014-12-12 11:17:16 -08:00
James Zern d456ccbc9d vp9_loopfilter_mmx: remove some unused tables
Change-Id: I964d25cc91c8e4864d73b142d9c7a1b39cb6cfbb
2014-12-12 11:16:24 -08:00
Jim Bankoski d916b0f22f Merge "vp9_dx_iface.c uses CONFIG_VP9_POSTPROC but config.h not included" 2014-12-12 11:10:17 -08:00
Jim Bankoski cf6a5c8759 Merge "Adds a test to make sure encoder parms get to decoder." 2014-12-12 11:10:08 -08:00
James Zern d5484f109e x86_abi_support: set LIBVPX_RAND w/vp9-postproc
set LIBVPX_RAND with --enable-vp9-postproc, previously only the vp8
config was checked. this fixes the build with --disable-postproc.

Change-Id: Ia61baded6aa0e44d6443ae4a3c85915f1054f053
2014-12-12 11:07:29 -08:00
Jingning Han 3e0793b80b Merge "Fix PICK_MODE_CONTEXT index in non-RD coding mode" 2014-12-12 09:16:01 -08:00
Jim Bankoski c67859f737 vp9_dx_iface.c uses CONFIG_VP9_POSTPROC but config.h not included
Change-Id: Id316b3786214bf1028992968955da917e3f2d4a3
2014-12-12 08:42:36 -08:00
Jim Bankoski 18d323606d Fix test to call clear system state in convolve_test.
Assembly tests should clear system state, as we have no
expectation of proper system state in between test runs..

Change-Id: I0f591996c1f17ef2a5a8572a6b445f757223a144
2014-12-12 06:18:56 -08:00
Jim Bankoski a0d9a9dd51 Adds a test to make sure encoder parms get to decoder.
This is meant as a framework for testing that encode parms make it
through to the decoder.

Change-Id: Idb86ee3668b45b4e73c23c6e4daef94b0650b786
2014-12-11 17:34:32 -08:00
Jingning Han e2c2a65695 Fix PICK_MODE_CONTEXT index in non-RD coding mode
This commit fixes a bug in the PICK_MODE_CONTEXT index for
horizontal partition case. The compression performance change
is less than 0.01% level, since most blocks are selected to
use square block size in RTC coding mode.

Change-Id: I67effc18ae8795fccdd82a55f4efc609fa5cb3e1
2014-12-11 17:21:24 -08:00
JackyChen 3425d6c83e Merge "Multiframe Quality Enhancement(MFQE) in VP9." 2014-12-11 16:24:08 -08:00
Marco 7e99cd2a9b Allow for 4x4 prediction blocks for key frame, speed 6.
For key frame under variance source partition: 4x4 prediction blocks
may be selected when variance of 8x8 block is very high (threshold is set fairly high for now).

Testing on some RTC clips shows this helps to reduce some ringing artifacts on key frame.
Encoded key frame size increases about ~10%. Key frame PSNR increases about ~0.1-0.2dB.

Change-Id: I56e203fac32ea6ef69897fb3ea269c59cb50d174
2014-12-11 15:36:16 -08:00
Jingning Han 811c74cdfa Merge "Replace division with bit shift in choose_partitioning" 2014-12-11 13:30:03 -08:00
Deb Mukherjee 847636fe88 Merge "Re-enable 8x8 fdct/fht tests by changing tolerance" 2014-12-11 12:29:06 -08:00