diff options
| author | Rémi Denis-Courmont <remi@remlab.net> | 2023-11-19 13:24:29 +0200 |
|---|---|---|
| committer | Rémi Denis-Courmont <remi@remlab.net> | 2023-11-23 18:57:18 +0200 |
| commit | 3a134e82994ff49b784056d2dfce0230a8256ebd (patch) | |
| tree | 6f1c8083d50e9b8f089406fe5aa0c19614f99e2a /tests/checkasm/alacdsp.c | |
| parent | 4adb93dff05dd947878c67784d98c9a4e13b57a7 (diff) | |
| download | ffmpeg-3a134e82994ff49b784056d2dfce0230a8256ebd.tar.gz | |
lavu/fixed_dsp: optimise R-V V fmul_reverse
Gathers are (unsurprisingly) a notable exception to the rule that R-V V
gets faster with larger group multipliers. So roll the function to speed
it up.
Before:
vector_fmul_reverse_fixed_c: 2840.7
vector_fmul_reverse_fixed_rvv_i32: 2430.2
After:
vector_fmul_reverse_fixed_c: 2841.0
vector_fmul_reverse_fixed_rvv_i32: 962.2
It might be possible to further optimise the function by moving the
reverse-subtract out of the loop and adding ad-hoc tail handling.
Diffstat (limited to 'tests/checkasm/alacdsp.c')
0 files changed, 0 insertions, 0 deletions