lavu/fixed_dsp: R-V V vector_fmul_reverse

libavutil/riscv/fixed_dsp_init.c

@@ -25,6 +25,8 @@
 #include "libavutil/cpu.h"
 #include "libavutil/fixed_dsp.h"
 
+void ff_vector_fmul_reverse_fixed_rvv(int *dst, const int *src0,
+                                      const int *src1, int len);
 void ff_vector_fmul_add_fixed_rvv(int *dst, const int *src0, const int *src1,
                                   const int *src2, int len);
 int ff_scalarproduct_fixed_rvv(const int *v1, const int *v2, int len);
@@ -36,6 +38,7 @@ av_cold void ff_fixed_dsp_init_riscv(AVFixedDSPContext *fdsp)
     int flags = av_get_cpu_flags();
 
     if ((flags & AV_CPU_FLAG_RVV_I32) && (flags & AV_CPU_FLAG_RVB_ADDR)) {
+        fdsp->vector_fmul_reverse = ff_vector_fmul_reverse_fixed_rvv;
         fdsp->vector_fmul_add = ff_vector_fmul_add_fixed_rvv;
 
         if (flags & AV_CPU_FLAG_RVV_I64)

libavutil/riscv/fixed_dsp_rvv.S

@@ -20,6 +20,30 @@
 
 #include "asm.S"
 
+func ff_vector_fmul_reverse_fixed_rvv, zve32x
+        csrwi   vxrm, 0
+        vsetvli t0, zero, e16, m4, ta, ma
+        sh2add  a2, a3, a2
+        vid.v   v0
+        vadd.vi v0, v0, 1
+1:
+        vsetvli t0, a3, e16, m4, ta, ma
+        slli    t1, t0, 2
+        vrsub.vx v4, v0, t0 // v4[i] = [VL-1, VL-2... 1, 0]
+        sub     a2, a2, t1
+        vsetvli zero, zero, e32, m8, ta, ma
+        vle32.v v8, (a2)
+        sub     a3, a3, t0
+        vle32.v v16, (a1)
+        add     a1, a1, t1
+        vrgatherei16.vv v24, v8, v4 // v24 = reverse(v8)
+        vsmul.vv v16, v16, v24
+        vse32.v v16, (a0)
+        add     a0, a0, t1
+        bnez    a3, 1b
+
+        ret
+endfunc
 
 func ff_vector_fmul_add_fixed_rvv, zve32x
         csrwi   vxrm, 0