Commit Graph

716 Commits

Author SHA1 Message Date
Kaustubh Raste
ecc5998bcf Fix mips dspr2 build warning
Change-Id: Ia8fb3ed124f01384e7896e309c9ff22c05b40719
2016-11-22 17:49:17 +05:30
James Zern
ba016b710a Merge "*ppflags.h: remove unused *_DEBUG_* enum values" 2016-11-10 20:53:22 +00:00
James Zern
4807f1584c *ppflags.h: remove unused *_DEBUG_* enum values
usage of the vp8 versions was removed in:
3f72509 vp8: remove VP8_SET_DBG* control support

vp9 had the usage stripped even earlier.

Change-Id: I978142eb6492552cd29c9c6feb1e89acfc5f7b84
2016-11-08 21:09:16 -08:00
Johann
f5141ea45f Refine vp8_refining_search_sadx4 targeting
This uses the same sdx4df pointers as vp8_diamond_search_sadx4 and
should therefore target the same optimizations.

See e4ddf9db6a

Change-Id: Ic298e9b25c34bbe6b7a0799509355b0addb56675
2016-11-08 15:22:44 -08:00
Marco
ff38b8dfae vp8: Add control for gf boost for 1 pass cbr.
Control already exists for vp9, adding it to vp8.
Usage is only when error_resilient is off.
Added a datarate unittest for non-zero boost.

Change-Id: I4296055ebe2f4f048e8210f344531f6486ac9e35
2016-10-19 09:43:53 -07:00
Kaustubh Raste
56b1be1889 Merge "Optimize vp8 loopfilter msa functions" 2016-10-12 05:44:11 +00:00
Kaustubh Raste
06a6b28d75 Optimize vp8 loopfilter msa functions
Updated code to process in 8bit as saturation/clipping takes care of overflow

Change-Id: I35fb2c0e702fd91309cc391c5a7745a3b619a64c
2016-10-07 15:48:31 +05:30
Linfeng Zhang
7aa27bd62f [vpx highbd lpf NEON 1/6] horizontal 4
BUG=webm:1300

Change-Id: Idf441806e6bf397ff5ecd8776146b3f781f50c40
2016-10-06 14:03:04 -07:00
Kaustubh Raste
68f6f6c4cc Modify vp8 idct msa functions store method
vp8_short_inv_walsh4x4_msa - Optimized to process in short vector type
Updated below functions to store exact number of bytes in output rather than complete vector
idct4x4_addblk_msa
idct4x4_addconst_msa
dequant_idct4x4_addblk_msa
dequant_idct4x4_addblk_2x_msa
dequant_idct_addconst_2x_msa

Change-Id: Ic1b3752e2421dc7d70a082dcdaab9d140d7e5d9c
2016-10-05 10:12:12 +05:30
James Zern
1396d12103 *idct*_neon.c: add missing rtcd include
+ correct declarations as necessary

BUG=webm:1294

Change-Id: I719602df9a56e79188a78e7f8b31257c6d3cc11d
2016-09-30 11:41:26 -07:00
Johann
721354fe7f vp8: remove mmx functions
When they have sse2 equivalents.

Change-Id: I158f631a3bcecba57b36093ac10114b1904767a7
2016-09-29 15:25:27 -07:00
Johann
2663b092ae Rename _xmm functions to _sse2
Avoid the extra level of indirection/confusion.

Change-Id: I0555f639d67835df9fb7dac0c75085e9954805f1
2016-09-29 15:23:11 -07:00
Johann
1364cb58b4 Remove vp8_clear_system_state
Use vpx_clear_system_state instead.

Change-Id: Ia3e9122f69a2c690ddd7c7bc54f92ccb9ec18b3e
2016-09-29 13:22:49 -07:00
Johann
c7f9d0719d vp8: clean up rtcd
Remove lines which specify the same name for a function.

Change-Id: I956bd8ce2b81a2a8feab5621d28bd2499c2b4c2d
2016-09-29 12:10:01 -07:00
Johann
e4ddf9db6a Hook up vp8_diamond_search_sad_sse3
The original commit never set any 'specialize' line:
61311e6103

It appears the sadx4 version of function uses sdx4df calls to speed up
the search. There are no sse3 versions of the sdx4df functions, but
there are sse2 and msa versions.

There is a neon version of vpx_sad16x16x4d but not any of the smaller
versions. Perhaps if they existed this function could be expanded to use
them.

Change-Id: I936d7d6b1a3ff6dcd5a4d2322272708c47cdec13
2016-09-27 15:31:49 -07:00
Johann Koenig
b165451ad5 Merge "Un-Revert "Restore vp8_sixtap_predict4x4_neon"" 2016-09-26 19:11:00 +00:00
Johann
ab0e7a237a Use shifted value for sinpi8sqrt2
The value 35468 changes sign when stored in int16_t:
implicit conversion from 'int' to 'int16_t' (aka 'short')
changes value from 35468 to -30068

This negation requires adding back the original value to compensate.
Shifting the value keeps the value positive and saves a post-vqdmulh
shift.

This technique is used in webp and idct_dequant_full_2x_neon

BUG=b/28027557

Change-Id: I0c5ce09bea170fe08061856c2af6f841a557e0c3
2016-09-23 17:04:18 -07:00
Johann
1d14e42df7 Un-Revert "Restore vp8_sixtap_predict4x4_neon"
This restores d9dce2f48e

Switched to using signed shift-and-narrow. Instead of saturating
negative results to 0, it was saturating them to 255.

BUG=webm:817
BUG=webm:1273

Change-Id: I571095336aa4182e3288b17924fcaaece42b0a49
2016-09-23 14:58:57 -07:00
James Zern
fcf281b6a1 Merge "vp8: remove VP8_SET_DBG* control support" 2016-09-22 00:43:35 +00:00
Johann
2bed8b6acd Keep vp8 sixtap read within bounds
When filtering it needs 6 pixels: 2 prior to the source, the source, and
3 after the source.

When filtering 16 wide, that means 21. To accomplish this the SSE2 reads
[-2] to [5], [6] to [13], and [14] to [21], a total of 24 bytes (reading
in groups of 8 is easy)

The filter then shifts this last set to the top half of the register and
uses 'or' to combine it with the previous set.

Valgrind detected an issue reading pixels [19], [20] and [21]:
Address 0x7f581c2 is 434 bytes inside a block of size 441 alloc'd

Note: we only need pixels [16], [17], and [18] as context for [15].

To fix this, it now reads 8 bytes starting at [11], which re-loads [11]
through [13], but stops at [18] and does not over-read any values.

This is shifted by 5 and 'or'd with xmm1. Although the lower bits are
not cleared, they overlap directly with [11] through [13], so 'or'
produces the correct results.

Change-Id: I0c89c03afa660fc9b0108ac055d7bd403e493320
2016-09-21 16:17:07 -07:00
James Zern
3f72509587 vp8: remove VP8_SET_DBG* control support
the --enable-postproc-visualizer configure option remains as a no-op as
do the control names and values for compatibility
+ remove the corresponding debug flags from vpxdec: --pp-*

Change-Id: I4a001cd9962b59560d7d6bda6272d4ff32b8d37c
2016-09-20 20:19:36 -07:00
James Zern
b6e686b1ea Merge changes from topic 'Wshorten'
* changes:
  vp8: convert some uses of unsigned long to size_t
  vp8/encoder: quiet some -Wshorten-64-to-32 warnings
2016-09-20 23:17:20 +00:00
Johann Koenig
8478f97105 Merge "Enable ssse3 bilinear tests" 2016-09-20 21:46:50 +00:00
James Zern
8281da74b9 vp8: convert some uses of unsigned long to size_t
similar to changes that were done in vp9 for encoded frame size
reporting. has the side-effect of quieting a -Wshorten-64-to-32 warning.

Change-Id: I89f74cb617fc29334ee351dc8dfaa3b8cfd4e5af
2016-09-19 18:35:59 -07:00
James Zern
7a9e476072 Merge changes from topic 'clang-format'
* changes:
  apply clang-format
  .clang-format: update to 3.8.1
2016-09-16 07:11:33 +00:00
Johann
e813c2b416 Enable ssse3 bilinear tests
The code only has issues when xoffset == 0 and yoffset == 0 which
represents a simple copy. Presumably this case does not need to be
handled because the issue has existed since 2010.

BUG=webm:1287

Change-Id: Ic47e2653f3b729e99b40e53d8d2d8d1501edaaa9
2016-09-15 23:16:26 -07:00
James Zern
6ae58fd55e Merge "Revert "Restore vp8_sixtap_predict4x4_neon"" 2016-09-16 06:13:42 +00:00
Johann Koenig
7795e99296 Revert "Restore vp8_sixtap_predict4x4_neon"
This reverts commit d9dce2f48e.

Appears to be failing the SixtapPredict tests in some configurations and possibly test vectors as well.

Change-Id: Ica6aa83ebac47d0a76e451846e7da67b1c17a7d7
2016-09-16 06:12:49 +00:00
Johann Koenig
fdbe249991 Merge "Restore vp8_bilinear_predict4x4_neon" 2016-09-16 05:33:50 +00:00
Johann
43743b1d3e Restore vp8_bilinear_predict4x4_neon
This function was removed when clang started introducing alignment hints
which caused the 32 bit vld1_lane_u32/vst1_lane_u32 to fail:
https://llvm.org/bugs/show_bug.cgi?id=24421

The load has been rendered safe with an implementation ~indiscernible
performance-wise that uses _u8 and over-reads just a touch.

It is still ~5x faster than C in the unaligned case and doing both
filters.

BUG=webm:892
BUG=webm:1273

Change-Id: Icf7167189391b46202f47233bb585c24c42bcc36
2016-09-15 21:16:11 -07:00
Johann Koenig
7bc0733c27 Merge "Restore vp8_sixtap_predict4x4_neon" 2016-09-16 04:12:08 +00:00
Johann
1d2aaf58dd vp8 postproc: expand CONFIG_POSTPROC guard
postproc.c is overloaded and used for both postproc and internal stats.
If only --enable-internal-stats is specified there are issues with
non-existent struct members and unused functions.

Change-Id: I82367f1ffce659c3918c9f964dbce94a716fbb89
2016-09-16 03:52:19 +00:00
clang-format
5f6d143b41 apply clang-format
Change-Id: I501597b7c1e0f0c7ae2aea3ee8073f0a641b3487
2016-09-15 15:07:53 -07:00
Johann
d9dce2f48e Restore vp8_sixtap_predict4x4_neon
This function was removed when clang started introducing alignment hints
which caused the 32 bit vld1_lane_u32/vst1_lane_u32 to fail:
https://llvm.org/bugs/show_bug.cgi?id=24421

The load has been rendered safe with an implementation ~indiscernible
performance-wise that uses _u8 and over-reads just a touch.

The store, when unaligned, has a version that is ~25% slower but safe
when xoffset = 0 (second pass filter only). When the first pass filter
(or both) are in play, the new version is almost identical in speed.

Worst case performance (both filters, unaligned stores) is roughly 3-4x
faster than C.

BUG=webm:817
BUG=webm:1273

Change-Id: I1e490e94453e0872151fe0dafb05557463f6247d
2016-09-15 14:56:47 -07:00
James Zern
a22a455899 cosmetics,vp8: join some lines, fix table format
Change-Id: Idcf3b68f0e59bd74c9d332bbd4a7c1484ddb691a
2016-09-09 16:39:34 -07:00
James Zern
d6d3d4ba31 get_cpu_count: quiet -Wshorten-64-to-32 warnings
sysconf returns a long; cast (unsigned) dwNumberOfProcessors to int for
good measure

Change-Id: I1f181d7bd9a060c0898db41f66a5065394afdc4e
2016-09-01 19:54:00 -07:00
Johann
18b6691105 Remove CONFIG_DEBUG guards from assert()
When 'NDEBUG' is set, assert() generates no code.

Change-Id: Icf61cfc1a8f6e5f0770b3626d8c73ae968df1108
2016-08-31 23:01:57 -07:00
Aleksey Vasenev
343b6b09a1 Align thread entry point stack
_beginthreadex does not align the stack on 16-byte boundary as expected
by gcc.

On x86 targets, the force_align_arg_pointer attribute may be applied to
individual function definitions, generating an alternate prologue and
epilogue that realigns the run-time stack if necessary. This supports
mixing legacy codes that run with a 4-byte aligned stack with modern
codes that keep a 16-byte stack for SSE compatibility.
https://gcc.gnu.org/onlinedocs/gcc/x86-Function-Attributes.html

Change-Id: Ie4e4ab32948c238fa87054d5664189972ca6708e
Signed-off-by: Aleksey Vasenev <margtu-fivt@ya.ru>
2016-08-10 11:57:34 -07:00
Johann
d55724fae9 Remove armv6 target
Change-Id: I1fa81cc9cabf362a185fc3a53f1e58de533a41e5
2016-08-04 12:55:06 -07:00
Johann
a7a8e07a44 Pad 'Left' when building under ASan
The neon intrinsics are not able to load just the 4 values that are
used. In vpx_dsp/arm/intrapred_neon.c:dc_4x4 it loads 8 values for both
the 'above' and 'left' computations, but only uses the sum of the first
4 values.

BUG=webm:1268

Change-Id: I937113d7e3a21e25bebde3593de0446bf6b0115a
2016-08-03 16:38:51 -07:00
clang-format
c42d54c3a3 vp8/postproc.c: disable clang-format for RGB_TO_YUV
Change-Id: Id2a936301ec1e3d5648b4f8adbf4e6625002589d
2016-07-23 10:55:44 -07:00
James Zern
b2542417cd vp8/postproc: fix implicit float conversion
float->int as reported by -Wfloat-conversion

Change-Id: I0089e8847b218c47526bcfbb0fffd9aad7c5adb3
2016-07-22 16:01:52 -07:00
Yunqing Wang
4b073bc39a Add back header in threading.h
Added back the header needed in threading.h

Change-Id: I2ce66ad4fe58004997623f6c3f3b8dd11640aa98
2016-07-21 17:26:05 -07:00
Yunqing Wang
87c6c5224d Revert "Amend and improve VP8 multithreading implementation"
Reverted the patch because of possible performance issue.

Change-Id: I49944f827ccd38ed194c9f8d9cb9036fa9bf79e1
2016-07-21 12:28:25 -07:00
James Zern
96797e43b4 vp8: remove extra semicolons
Change-Id: I84e1a293ee033865f82c244e8aaaadfb2fb27e63
2016-07-19 20:44:14 -07:00
clang-tidy
7f3e07f1c8 vp8: apply clang-tidy google-readability-braces-around-statements
applied against an x86_64 configure

clang-tidy-3.7.1 \
  -checks='-*,google-readability-braces-around-statements' \
  -header-filter='.*' -fix
+ clang-format afterward

Change-Id: I6694edeaee89b58b8b3082187e6756561136b459
2016-07-19 12:38:03 -07:00
Jim Bankoski
3e04114f3d prepend ++ instead of post in for loops.
Applied the following regex  :
search for: (for.*\(.*;.*;) ([a-zA-Z_]*)\+\+\)
replace with: \1 ++\2)

This misses some for loops:
ie : for (mb_col = 0; mb_col < oci->mb_cols; mb_col++, mi++)

Change-Id: Icf5f6fb93cced0992e0bb71d2241780f7fb1f0a8
2016-07-18 06:54:50 -07:00
clang-format
81a6739533 vp8: apply clang-format
Change-Id: I7605b6678014a5426ceb45c27b54885e0c4e06ed
2016-07-15 19:28:44 -07:00
Jim Bankoski
cb957c302a addnoise : clear out static size for generated noise
Change-Id: I5d4343f2da9cd4b01dd37be7a048d159fec109d1
2016-07-15 15:52:45 -07:00
Jim Bankoski
0dc69c70f7 postproc : fix function parameters for noise functions.
Change-Id: I582b6307f28bfc987dcf8910379a52c6f679173c
2016-07-15 08:27:34 -07:00