virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2026-06-19 19:03:00 +02:00
Code Issues Releases Activity
110,895 Commits 39 Branches 425 Tags
36c16a06b2e31a39c236d724c2fd5d457f02493a
Commit Graph
63 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
James Darnley 073ec3b9da avfilter/bwdif: add avx2 filter_line function 
8-bit:
2.24x faster (1925±1.3 vs. 859±2.2 decicycles) compared with ssse3
10-bit:
2.00x faster (1703±1.7 vs. 853±2.0 decicycles) compared with ssse3
 2023-03-25 02:38:17 +01:00
James Darnley a937723ca9 tests: add bwdif to fate filter tests 2023-03-25 02:38:17 +01:00
James Darnley 087faf8cac checkasm: add test for bwdif 2023-03-25 02:38:17 +01:00
James Darnley b503b5a0cf avfilter/bwdif: move filter_line init to a dedicated function 2023-03-25 02:38:17 +01:00
James Darnley e50a02b0f6 tests: actually test yadif's 10 and 16-bit functions 2023-03-06 21:41:24 +01:00
James Darnley eef763c705 checkasm/v210dec: add extra space to the destination arrays 2022-12-21 00:36:49 +01:00
James Darnley 6af453ca38 avcodec/x86: add avx512icl function for v210dec 
Ice Lake (Xeon Silver 4316): 2.01x faster (1147±36.8 vs. 571±38.2 decicycles) compared with avx2
 2022-12-20 15:02:45 +01:00
James Darnley f30b4c2f47 avcodec/x86/v210: add some comments to the improved avx2 function 2022-12-20 15:02:45 +01:00
James Darnley f1f41d6ff5 configure: support lsan as toolchain 2022-12-20 15:02:45 +01:00
James Darnley 5dfb4f9690 avcodec/x86/v210enc: change '0b' binary constant prefix to 'b' suffix 
For compatability with yasm from 0.7.0
 2022-12-03 16:44:24 +01:00
James Darnley 690b7890f0 avcodec/x86/v210enc: remove unneeded instruction 2022-12-01 18:19:03 +01:00
James Darnley c67a2b14a2 avcodec/x86/v210enc: expand and correct comments 2022-12-01 18:19:03 +01:00
James Darnley 651cb867b1 avcodec/v210enc: add new 10-bit function for avx512 avx512icl 
avx512 on Skylake-X (Xeon D-2123IT):
1.19x faster (970±91.2 vs. 817±104.4 decicycles) compared with avx2

avx512icl on Ice Lake (Xeon Silver 4316):
2.52x faster (1350±5.3 vs. 535±9.5 decicycles) compared with avx2
 2022-12-01 18:19:03 +01:00
James Darnley bda53d2dde avcodec/x86/v210enc: replace register use with named register 2022-12-01 18:19:03 +01:00
James Darnley cfd1c3c0a1 checkasm/v210enc: test the entire width of 10-bit planar input arrays 2022-12-01 18:19:03 +01:00
James Darnley c3d36e1b3d avcodec/v210enc: add new function for avx2 avx512 avx512icl 
Negligible speed difference for avx2 on Zen 2 (Ryzen 5700X) and
Broadwell (Xeon E5-2620 v4):
    1690±4.3 decicycles vs. 1693±78.4
    1439±31.1 decicycles vs 1429±16.7

Moderate speedup with avx512 on Skylake-X (Xeon D-2123IT):
1.22x faster (793±0.8 vs. 649±5.5 decicycles) compared with avx2

Better speedup with avx512icl on Ice Lake (Xeon Silver 4316):
1.77x faster (784±1.8 vs. 442±11.6 decicycles) compared with avx2

Co-authors:
    Henrik Gramner <henrik@gramner.com>
    Kieran Kunhya <kierank@obe.tv>
 2022-11-04 19:37:46 +01:00
James Darnley 1936c06f02 checkasm: add a verbose check function for uint32_t data 2022-11-04 19:37:46 +01:00
James Darnley 0f252dfa95 avutil/tests/cpu: print the avx512icl flag 2022-11-04 19:37:46 +01:00
James Darnley f53c590f3d mailmap: stop git lying about who I commit things as 2022-11-04 19:37:46 +01:00
James Darnley 46f1718cd9 avcodec/x86/v210: fix operands of vpblendd used in new avx2 code 
Assembly failed when using yasm rather than nasm.
 2019-05-02 21:20:54 +02:00
James Darnley 76c370af64 checkasm: add test for v210dec 2019-05-02 19:21:37 +02:00
James Darnley e2cbf24f27 avcodec/v210dec: move DSP function setting into dedicated function 
Prepare for checkasm test.
 2019-05-02 19:21:37 +02:00
James Darnley 40d4b13228 checkasm: support for AVX-512 functions 2017-12-24 22:02:41 +01:00
James Darnley 8f86e66238 avcodec: add stride alignment needed for AVX-512 2017-12-24 22:02:41 +01:00
James Darnley e2218ed8ce avutil: add alignment needed for AVX-512 2017-12-24 22:02:41 +01:00
James Darnley 4783a01c11 avutil: detect when AVX-512 is available 2017-12-24 22:02:41 +01:00
James Darnley 8b81eabe57 avutil: add AVX-512 flags 2017-12-24 22:02:41 +01:00
James Darnley b7a3d1f249 configure: test whether x86 assembler supports AVX-512 2017-12-24 22:02:41 +01:00
James Darnley 0c2acccd4b avcodec/x86: use new x86-64 functions for -idct simple 
They now match according to FATE, barring any further bugs with untested
parts
 2017-06-28 17:27:35 +02:00
James Darnley d7246ea9f2 avcodec/x86: add an 8-bit simple IDCT function based on the x86-64 high depth functions 
Includes add/put functions

Rounding contributed by Ronald S. Bultje
 2017-06-28 17:27:35 +02:00
James Darnley 8b19467d07 avcodec/x86: allow future 8-bit simple idct to have "DC only hack" 
Created by Ronald S. Bultje
 2017-06-28 17:27:35 +02:00
James Darnley 8221c71703 avcodec/x86: allow future 8-bit simple idct to use slightly different coefficients 2017-06-20 16:12:25 +02:00
James Darnley 9d11fedd11 avcodec/mdec: override IDCT choice before initing DSP structs 2017-06-20 13:59:51 +02:00
James Darnley d2597fb0c1 avcodec/x86: modify simple_idct10 macros to add an action paramter 2017-06-20 13:35:01 +02:00
James Darnley 8781330d80 avcodec/x86: cleanup simple_idct10 
Use named arguments for the functions so we can remove a define.  The
stride/linesize argument is now ptrdiff_t type so we no longer need to
sign extend the register.
 2017-06-20 13:34:38 +02:00
James Darnley e3db94302c avcodec/x86/mpegenc: support transpose permuation type 2017-06-20 12:12:13 +02:00
James Darnley fa30a0a548 avcodec/x86/mpegenc: check IDCT permutation type is a valid value 2017-06-20 12:12:13 +02:00
James Darnley 0dea0114fb avcodec/x86/idctdsp_init: reindent 2017-05-30 13:20:44 +02:00
James Darnley 8e89f6fd37 avcodec/x86: move simple_idct to external assembly 2017-05-30 13:20:42 +02:00
James Darnley 7aa90b4e94 avcodec/h264: add sse2 versions of previous idct functions 
Kaby Lake Pentium:
 - ff_h264_idct_add_8_sse2:    ~1.18x faster than mmxext
 - ff_h264_idct_dc_add_8_sse2: ~1.07x faster than mmxext
 2017-05-15 15:00:20 +02:00
James Darnley 27460dfebc avcodec/h264: add avx 8-bit h264_idct_dc_add 
Haswell:
 - 1.02x faster (405±0.7 vs. 397±0.8 decicycles) compared with mmxext

Skylake-U:
 - 1.06x faster (498±1.8 vs. 470±1.3 decicycles) compared with mmxext
 2017-05-15 15:00:19 +02:00
James Darnley f61d454ca1 avcodec/h264: add avx 8-bit h264_idct_add 
Haswell:
 - 1.11x faster (522±0.4 vs. 469±1.8 decicycles) compared with mmxext

Skylake-U:
 - 1.21x faster (671±5.5 vs. 555±1.4 decicycles) compared with mmxext
 2017-05-15 15:00:17 +02:00
James Darnley b5325c6711 avcodec/h264: use some 3 operand forms 2017-05-15 15:00:16 +02:00
James Darnley 060ba9e5e3 avcodec/h264: change RETs into REP_RETs where appropriate 2017-05-15 15:00:15 +02:00
James Darnley 33de0fee2c avcodec/h264: enable sse2 chroma deblock/loop filter functions 
Between 1.00 and 1.16 times faster on Intel Yorkfield Core 2 Quad.
Between 1.11 and 1.39 times faster on Intel Kaby Lake Pentium.
 2017-02-27 13:22:06 +01:00
James Darnley cd893b9307 avcodec/h264: add avx 8-bit 4:2:2 chroma h intra deblock/loop filter 
~1.37x faster (147 vs. 108 cycles) compared to mmxext function
 2017-02-27 13:22:06 +01:00
James Darnley 0e16b3e2be avcodec/h264: add avx 8-bit 4:2:0 chroma h intra deblock/loop filter 
~1.10x faster (69 vs. 63 cycles) compared to mmxext function
 2017-02-27 13:22:06 +01:00
James Darnley 987ffe4b8d avcodec/h264: add avx 8-bit chroma v intra deblock/loop filter 
~1.14x faster (90 vs 78 cycles) compared with mmxext
 2017-02-27 13:22:06 +01:00
James Darnley 88307b3eec avcodec/h264: add avx 8-bit 4:2:2 chroma h deblock/loop filter 
~1.21x faster (68 vs. 56 cycles) compared with mmxext function
 2017-02-27 13:22:06 +01:00
James Darnley ac096fc82d avcodec/h264: add avx 8-bit 4:2:0 chroma h deblock/loop filter 
~1.14x faster (93 vs. 81 cycles) compared with mmxext function
 2017-02-27 13:22:06 +01:00
James Darnley 5c56758843 avcodec/h264: add avx 8-bit chroma v deblock/loop filter 
~1.24x faster (101 vs. 81 cycles) compared with mmxext function
 2017-02-27 13:22:06 +01:00
James Darnley 5336887867 avcodec/h264: sse2, avx h luma mbaff deblock/loop filter 
x86-64 only

Yorkfield:
- sse2: ~2.17x (434 vs. 200 cycles)

Nehalem:
- sse2: ~2.94x (409 vs. 139 cycles)

Skylake:
- sse2: ~3.10x (370 vs. 119 cycles)
- avx:  ~3.29x (370 vs. 112 cycles)
 2017-02-18 20:26:52 +01:00
James Darnley 7627df15d4 x86util: import MOVHL macro 
Originally committed to x264 in 1637239a by Henrik Gramner who has
agreed to re-license it as LGPL.  Original commit message follows.

    x86: Avoid some bypass delays and false dependencies

    A bypass delay of 1-3 clock cycles may occur on some CPUs when transitioning
    between int and float domains, so try to avoid that if possible.
 2017-02-18 20:26:51 +01:00
James Darnley e18bc2114f avcodec/h264: add named parameters to x86 function 2017-02-18 20:26:50 +01:00
James Darnley 9d815b7424 avcodec/x86: deduplicate PASS8ROWS macro 2017-02-18 20:26:49 +01:00
James Darnley acdd2d805d avcodec/h264: resolve assert being triggered when stack is not aligned 
32-bit msvc.
 2016-12-07 22:32:19 +01:00
James Darnley 728651df06 avcodec/h264: mmx2, sse2, avx 10-bit 4:2:2 h chroma deblock/loop filter 
Yorkfield:
 - mmx2: 2.53x (504 vs. 199 cycles)
 - sse2: 3.83x (504 vs. 131 cycles)

Nehalem:
 - mmx2: 2.42x (365 vs. 151 cycles)
 - sse2: 3.56x (365 vs. 103 cycles)

Skylake:
 - mmx2: 1.81x (308 vs. 170 cycles)
 - sse2: 2.84x (308 vs. 108 cycles)
 - avx:  2.93x (308 vs. 105 cycles)
 2016-12-07 00:29:13 +01:00
James Darnley add21d0bb3 avcodec/h264: mmx2, sse2, avx 10-bit h chroma deblock/loop filter 
Yorkfield:
 - mmx2: 2.45x (279 vs. 114 cycles)
 - sse2: 3.36x (279 vs.  83 cycles)

Nehalem:
 - mmx2: 2.10x (192 vs.  92 cycles)
 - sse2: 2.84x (192 vs.  68 cycles)

Skylake:
 - mmx2: 1.75x (170 vs.  97 cycles)
 - sse2: 2.47x (170 vs.  69 cycles)
 - avx:  2.47x (170 vs.  69 cycles)
 2016-12-07 00:29:13 +01:00
James Darnley 58ca2ef62e whitespace changes after last commit 2016-12-07 00:29:13 +01:00
James Darnley f33714a694 avcodec/h264: clean up and expand x86 function definitions 2016-12-07 00:29:13 +01:00
James Darnley 13d71c28cc avcodec/h264: sse2 and avx 4:2:2 idct add8 10-bit functions 
Yorkfield:
 - sse2:
   - complex: 4.13x faster (1514 vs. 367 cycles)
   - simple:  4.38x faster (1836 vs. 419 cycles)

Skylake:
 - sse2:
   - complex: 3.61x faster ( 936 vs. 260 cycles)
   - simple:  3.97x faster (1126 vs. 284 cycles)
 - avx (versus sse2):
   - complex: 1.07x faster (260 vs. 244 cycles)
   - simple:  1.03x faster (284 vs. 274 cycles)
 2016-11-30 22:58:28 +01:00
James Darnley 1dae7ffa0b avcodec/h264: mmx 4:2:2 idct add8 function 
2.87 times faster (1830 vs. 638 cycles)
 2016-11-30 22:58:27 +01:00
James Darnley 815ea8c6cc avcodec/h264: mmxext 4:2:2 chroma intra deblock/loop filter 
2.1 times faster (401 vs. 194 cycles)
 2016-11-30 22:58:27 +01:00