diff options
| author | Rémi Denis-Courmont <remi@remlab.net> | 2023-07-17 20:11:08 +0300 |
|---|---|---|
| committer | Rémi Denis-Courmont <remi@remlab.net> | 2023-07-20 22:54:34 +0300 |
| commit | 29b9d616c212f20bc0d3e24928919a14efc047f8 (patch) | |
| tree | d3d2d303d587e40270500d90364368137ad3e96c /libavcodec/arm/pixblockdsp_neon.S | |
| parent | b710f881cefde728f869689193885b82af6b90cd (diff) | |
| download | ffmpeg-29b9d616c212f20bc0d3e24928919a14efc047f8.tar.gz | |
lavu/float_dsp: rework RISC-V V scalar product
1) Take the reductive sum out of the loop,
leaving a regular vector addition in the loop.
2) Merge the addition and the multiplication.
3) Unroll.
Before:
scalarproduct_float_rvv_f32: 832.5
After:
scalarproduct_float_rvv_f32: 275.2
Diffstat (limited to 'libavcodec/arm/pixblockdsp_neon.S')
0 files changed, 0 insertions, 0 deletions