mbox series

[FFmpeg-devel,v2,00/14] aarch64/vvc: Add SIMD

Message ID tencent_DA162A4D1FA026CDFFCFF655132715B89C07@qq.com
Headers show
Series aarch64/vvc: Add SIMD | expand

Message

Zhao Zhili Sept. 11, 2024, 6:06 p.m. UTC 
From: Zhao Zhili <zhilizhao@tencent.com>

Patches 1~9 has been updated according to Martin's review.

Patches 10~14 are new.

I have created a PR on github:
https://github.com/quink-black/FFmpeg/pull/2

Zhao Zhili (14):
  aarch64/hevc: Simplify function prototypes by macro
  aarch64/hevc: Move epel/qpel to h26x directory
  aarch64/vvc: Add put_qpel_h_* and put_qpel_uni_h_*
  aarch64/vvc: Add put_pel/put_pel_uni/put_pel_uni_w
  aarch64/vvc: Add put_qpel_hx i8mm
  avcodec/hevc: ff_hevc_(qpel/epel)_filters are signed type
  aarch64/h26x: Remove duplicate b.eq instruction
  aarch64/vvc: Add put_qpel_vx
  aarch64/vvc: Add put_qpel_hv
  aarch64/vvc: Add sad
  aarch64/vvc: Add put_epel_h
  aarch64/vvc: Add put_epel_h i8mm
  aarch64/vvc: Add put_epel_hv
  aarch64/vvc: Add avg

 libavcodec/aarch64/Makefile                   |   4 +-
 libavcodec/aarch64/h26x/dsp.h                 | 268 ++++++++
 .../{hevcdsp_epel_neon.S => h26x/epel_neon.S} | 404 +++++++++---
 .../{hevcdsp_qpel_neon.S => h26x/qpel_neon.S} | 592 ++++++++++++++++--
 libavcodec/aarch64/hevcdsp_init_aarch64.c     | 227 -------
 libavcodec/aarch64/vvc/Makefile               |   4 +
 libavcodec/aarch64/vvc/dsp_init.c             | 114 ++++
 libavcodec/aarch64/vvc/inter.S                | 163 +++++
 libavcodec/aarch64/vvc/sad.S                  |  75 +++
 libavcodec/hevc/dsp_template.c                |   4 +-
 10 files changed, 1480 insertions(+), 375 deletions(-)
 rename libavcodec/aarch64/{hevcdsp_epel_neon.S => h26x/epel_neon.S} (94%)
 rename libavcodec/aarch64/{hevcdsp_qpel_neon.S => h26x/qpel_neon.S} (91%)
 create mode 100644 libavcodec/aarch64/vvc/inter.S
 create mode 100644 libavcodec/aarch64/vvc/sad.S

Comments

Martin Storsjö Sept. 12, 2024, 10:50 a.m. UTC | #1 
On Thu, 12 Sep 2024, Zhao Zhili wrote:

> From: Zhao Zhili <zhilizhao@tencent.com>
>
> Patches 1~9 has been updated according to Martin's review.
>
> Patches 10~14 are new.
>
> I have created a PR on github:
> https://github.com/quink-black/FFmpeg/pull/2

Thanks for testing it through that set of tests!

No further comments from me on this set, it seems reasonable.

// Martin