Re: [FFmpeg-devel] [PATCH v4] libavfilter/x86/vf_convolution: add sobel filter optimization and unit test with intel AVX512 VNNI

["Wang, Bin" <bin.wang-at-intel.com@ffmpeg.org>]

Thanks for the review, based on the comments, patch v5 is sent out: https://patchwork.ffmpeg.org/project/ffmpeg/patch/20220920103305.7902-1-bin.wang@intel.com/



Changes:

  1.  Remove redundant zero-initializations in asm code
  2.  Pass depth and nb_planes as parameters to ff_sobel_init_x86()
  3.  Filter-agnostic initialization was moved to the beginning of param_init()
  4.  I also noticed the duplicate configurations. But I think it may not suitable to change other code path in a sobel patch, just change sobel code path.
  5.  In order to move sobel initialization to convolution.h, I have to move 3 functions to this header file. They are setup_3x3(), filter_sobel() and filter16_sobel()



Welcome more advice!



---

libavfilter/convolution.h             |  74 +++++++++++++

libavfilter/vf_convolution.c          |  91 +++-------------

libavfilter/x86/vf_convolution.asm    | 147 ++++++++++++++++++++++++++

libavfilter/x86/vf_convolution_init.c |  18 ++++

tests/checkasm/Makefile               |   1 +

tests/checkasm/checkasm.c             |   3 +

tests/checkasm/checkasm.h             |   1 +

tests/checkasm/vf_convolution.c       | 103 ++++++++++++++++++

8 files changed, 360 insertions(+), 78 deletions(-)  create mode 100644 tests/checkasm/vf_convolution.c







Thanks

Bin

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".