diff options
| author | Rémi Denis-Courmont <remi@remlab.net> | 2022-09-26 17:52:35 +0300 |
|---|---|---|
| committer | Lynne <dev@lynne.ee> | 2022-09-27 13:19:52 +0200 |
| commit | 47ce9735cc8362ea6b3a508dd660233af28d77ba (patch) | |
| tree | 5a29ca0c70815b3d691af829ec5088ad2762bb5c | |
| parent | f4ea45040f4c0d896fce66a6f52deb53249ed77f (diff) | |
| download | ffmpeg-47ce9735cc8362ea6b3a508dd660233af28d77ba.tar.gz | |
lavu/floatdsp: RISC-V V butterflies_float
| -rw-r--r-- | libavutil/riscv/float_dsp_init.c | 2 | ||||
| -rw-r--r-- | libavutil/riscv/float_dsp_rvv.S | 18 |
2 files changed, 20 insertions, 0 deletions
diff --git a/libavutil/riscv/float_dsp_init.c b/libavutil/riscv/float_dsp_init.c index 8982436647..a1cd180cdc 100644 --- a/libavutil/riscv/float_dsp_init.c +++ b/libavutil/riscv/float_dsp_init.c @@ -33,6 +33,7 @@ void ff_vector_fmul_scalar_rvv(float *dst, const float *src, float mul, int len); void ff_vector_fmul_add_rvv(float *dst, const float *src0, const float *src1, const float *src2, int len); +void ff_butterflies_float_rvv(float *v1, float *v2, int len); void ff_vector_dmul_rvv(double *dst, const double *src0, const double *src1, int len); @@ -51,6 +52,7 @@ av_cold void ff_float_dsp_init_riscv(AVFloatDSPContext *fdsp) fdsp->vector_fmac_scalar = ff_vector_fmac_scalar_rvv; fdsp->vector_fmul_scalar = ff_vector_fmul_scalar_rvv; fdsp->vector_fmul_add = ff_vector_fmul_add_rvv; + fdsp->butterflies_float = ff_butterflies_float_rvv; } if (flags & AV_CPU_FLAG_RVV_F64) { diff --git a/libavutil/riscv/float_dsp_rvv.S b/libavutil/riscv/float_dsp_rvv.S index db62402878..a721c44667 100644 --- a/libavutil/riscv/float_dsp_rvv.S +++ b/libavutil/riscv/float_dsp_rvv.S @@ -93,6 +93,24 @@ func ff_vector_fmul_add_rvv, zve32f ret endfunc +// (a0) = (a0) + (a1), (a1) = (a0) - (a1) [0..a2-1] +func ff_butterflies_float_rvv, zve32f +1: + vsetvli t0, a2, e32, m1, ta, ma + vle32.v v16, (a0) + sub a2, a2, t0 + vle32.v v24, (a1) + vfadd.vv v0, v16, v24 + vfsub.vv v8, v16, v24 + vse32.v v0, (a0) + sh2add a0, t0, a0 + vse32.v v8, (a1) + sh2add a1, t0, a1 + bnez a2, 1b + + ret +endfunc + // (a0) = (a1) * (a2) [0..a3-1] func ff_vector_dmul_rvv, zve64d 1: |