[FFmpeg-devel] [PATCH 18/31] lavu/floatdsp: RISC-V V scalarproduct_float

[remi at remlab.net]

From: Rémi Denis-Courmont <remi at remlab.net>

---
 libavutil/riscv/float_dsp_init.c |  2 ++
 libavutil/riscv/float_dsp_rvv.S  | 20 ++++++++++++++++++++
 2 files changed, 22 insertions(+)

diff --git a/libavutil/riscv/float_dsp_init.c b/libavutil/riscv/float_dsp_init.c
index dacd81c08b..cc9b7e83dc 100644
--- a/libavutil/riscv/float_dsp_init.c
+++ b/libavutil/riscv/float_dsp_init.c
@@ -38,6 +38,7 @@ void ff_vector_fmul_add_rvv(float *dst, const float *src0, const float *src1,
 void ff_vector_fmul_reverse_rvv(float *dst, const float *src0,
                                  const float *src1, int len);
 void ff_butterflies_float_rvv(float *v1, float *v2, int len);
+float ff_scalarproduct_float_rvv(const float *v1, const float *v2, int len);
 
 void ff_vector_dmul_rvv(double *dst, const double *src0, const double *src1,
                          int len);
@@ -59,6 +60,7 @@ av_cold void ff_float_dsp_init_riscv(AVFloatDSPContext *fdsp)
         fdsp->vector_fmul_add = ff_vector_fmul_add_rvv;
         fdsp->vector_fmul_reverse = ff_vector_fmul_reverse_rvv;
         fdsp->butterflies_float = ff_butterflies_float_rvv;
+        fdsp->scalarproduct_float = ff_scalarproduct_float_rvv;
 
         if (flags & AV_CPU_FLAG_RV_ZVE64D) {
             fdsp->vector_dmul = ff_vector_dmul_rvv;
diff --git a/libavutil/riscv/float_dsp_rvv.S b/libavutil/riscv/float_dsp_rvv.S
index ce530f6108..ab2e0c42d7 100644
--- a/libavutil/riscv/float_dsp_rvv.S
+++ b/libavutil/riscv/float_dsp_rvv.S
@@ -165,6 +165,26 @@ func ff_butterflies_float_rvv, zve32f
         ret
 endfunc
 
+// a0 = (a0).(a1) [0..a2-1]
+func ff_scalarproduct_float_rvv, zve32f
+        vsetvli      zero, zero, e32, m1, ta, ma
+        vmv.s.x      v8, zero
+1:
+        vsetvli      t0, a2, e32, m1, ta, ma
+        vle32.v      v16, (a0)
+        sub          a2, a2, t0
+        vle32.v      v24, (a1)
+        sh2add       a0, t0, a0
+        vfmul.vv     v16, v16, v24
+        sh2add       a1, t0, a1
+        vfredusum.vs v8, v16, v8
+        bnez         a2, 1b
+
+        vfmv.f.s fa0, v8
+NOHWF   fmv.x.w  a0, fa0
+        ret
+endfunc
+
 // (a0) = (a1) * (a2) [0..a3-1]
 func ff_vector_dmul_rvv, zve64d
 1:
-- 
2.37.2