[FFmpeg-devel] [PATCH 24/31] lavc/fmtconvert: RISC-V V int32_to_float_fmul_array8

[remi at remlab.net]

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

---
 libavcodec/riscv/fmtconvert_init.c |  7 ++++++-
 libavcodec/riscv/fmtconvert_rvv.S  | 28 ++++++++++++++++++++++++++++
 2 files changed, 34 insertions(+), 1 deletion(-)

diff --git a/libavcodec/riscv/fmtconvert_init.c b/libavcodec/riscv/fmtconvert_init.c
index fd2f58d060..1796717a1c 100644
--- a/libavcodec/riscv/fmtconvert_init.c
+++ b/libavcodec/riscv/fmtconvert_init.c
@@ -27,13 +27,18 @@
 
 void ff_int32_to_float_fmul_scalar_rvv(float *dst, const int32_t *src,
                                        float mul, int len);
+void ff_int32_to_float_fmul_array8_rvv(FmtConvertContext *c, float *dst,
+                                       const int32_t *src, const float *mul,
+                                       int len);
 
 av_cold void ff_fmt_convert_init_riscv(FmtConvertContext *c)
 {
 #ifdef HAVE_RVV
     int flags = av_get_cpu_flags();
 
-    if (flags & AV_CPU_FLAG_RV_ZVE32F)
+    if (flags & AV_CPU_FLAG_RV_ZVE32F) {
         c->int32_to_float_fmul_scalar = ff_int32_to_float_fmul_scalar_rvv;
+        c->int32_to_float_fmul_array8 = ff_int32_to_float_fmul_array8_rvv;
+    }
 #endif
 }
diff --git a/libavcodec/riscv/fmtconvert_rvv.S b/libavcodec/riscv/fmtconvert_rvv.S
index b7c78831a0..c79f80cc47 100644
--- a/libavcodec/riscv/fmtconvert_rvv.S
+++ b/libavcodec/riscv/fmtconvert_rvv.S
@@ -37,3 +37,31 @@ NOHWF   mv          a2, a3
 
         ret
 endfunc
+
+func ff_int32_to_float_fmul_array8_rvv, zve32f
+        srai        a4, a4, 3
+
+1:      vsetvli     t0, a4, e32, m1, ta, ma
+        vle32.v     v24, (a3)
+        slli        t2, t0, 2 + 3
+        vlseg8e32.v v16, (a2)
+        vsetvli     t3, zero, e32, m8, ta, ma
+        vfcvt.f.x.v v16, v16
+        vsetvli     zero, a4, e32, m1, ta, ma
+        vfmul.vv    v16, v16, v24
+        sub         a4, a4, t0
+        vfmul.vv    v17, v17, v24
+        sh2add      a3, t0, a3
+        vfmul.vv    v18, v18, v24
+        add         a2, a2, t2
+        vfmul.vv    v19, v19, v24
+        vfmul.vv    v20, v20, v24
+        vfmul.vv    v21, v21, v24
+        vfmul.vv    v22, v22, v24
+        vfmul.vv    v23, v23, v24
+        vsseg8e32.v v16, (a1)
+        add         a1, a1, t2
+        bnez        a4, 1b
+
+        ret
+endfunc
-- 
2.37.2