diff options
| author | Rémi Denis-Courmont <remi@remlab.net> | 2022-09-26 17:52:31 +0300 |
|---|---|---|
| committer | Lynne <dev@lynne.ee> | 2022-09-27 13:19:52 +0200 |
| commit | da169a210dbd33d5d34baaf3d188ca5388b1b267 (patch) | |
| tree | 7c407041f04debd5c43fef23b8f281854816802d /libavutil/riscv | |
| parent | 7058af9969b737adbb1cd302cf8fa5feb7bc9e2b (diff) | |
| download | ffmpeg-da169a210dbd33d5d34baaf3d188ca5388b1b267.tar.gz | |
lavu/floatdsp: RISC-V V vector_dmul
Diffstat (limited to 'libavutil/riscv')
| -rw-r--r-- | libavutil/riscv/float_dsp_init.c | 6 | ||||
| -rw-r--r-- | libavutil/riscv/float_dsp_rvv.S | 17 |
2 files changed, 22 insertions, 1 deletions
diff --git a/libavutil/riscv/float_dsp_init.c b/libavutil/riscv/float_dsp_init.c index 2482094ab4..29114dfb82 100644 --- a/libavutil/riscv/float_dsp_init.c +++ b/libavutil/riscv/float_dsp_init.c @@ -30,6 +30,8 @@ void ff_vector_fmul_rvv(float *dst, const float *src0, const float *src1, void ff_vector_fmul_scalar_rvv(float *dst, const float *src, float mul, int len); +void ff_vector_dmul_rvv(double *dst, const double *src0, const double *src1, + int len); void ff_vector_dmul_scalar_rvv(double *dst, const double *src, double mul, int len); @@ -43,7 +45,9 @@ av_cold void ff_float_dsp_init_riscv(AVFloatDSPContext *fdsp) fdsp->vector_fmul_scalar = ff_vector_fmul_scalar_rvv; } - if (flags & AV_CPU_FLAG_RVV_F64) + if (flags & AV_CPU_FLAG_RVV_F64) { + fdsp->vector_dmul = ff_vector_dmul_rvv; fdsp->vector_dmul_scalar = ff_vector_dmul_scalar_rvv; + } #endif } diff --git a/libavutil/riscv/float_dsp_rvv.S b/libavutil/riscv/float_dsp_rvv.S index 00fb7354bb..710e122444 100644 --- a/libavutil/riscv/float_dsp_rvv.S +++ b/libavutil/riscv/float_dsp_rvv.S @@ -55,6 +55,23 @@ NOHWF mv a2, a3 ret endfunc +// (a0) = (a1) * (a2) [0..a3-1] +func ff_vector_dmul_rvv, zve64d +1: + vsetvli t0, a3, e64, m1, ta, ma + vle64.v v16, (a1) + sub a3, a3, t0 + vle64.v v24, (a2) + sh3add a1, t0, a1 + vfmul.vv v16, v16, v24 + sh3add a2, t0, a2 + vse64.v v16, (a0) + sh3add a0, t0, a0 + bnez a3, 1b + + ret +endfunc + // (a0) = (a1) * fa0 [0..a2-1] func ff_vector_dmul_scalar_rvv, zve64d NOHWD fmv.d.x fa0, a2 |