[FFmpeg-cvslog] lavc/vc1dsp: fuse multiply-adds in R-V V inv_trans_4

[Rémi Denis-Courmont]

ffmpeg | branch: master | Rémi Denis-Courmont <remi at remlab.net> | Sun Jun 30 14:15:33 2024 +0300| [78e1565f847588cb3d34d1a920902fb253d9c03c] | committer: Rémi Denis-Courmont

lavc/vc1dsp: fuse multiply-adds in R-V V inv_trans_4

T-Head C908 (cycles):            before   after
vc1dsp.vc1_inv_trans_4x4_rvv_i32: 128.0   120.0
vc1dsp.vc1_inv_trans_4x8_rvv_i32: 244.0   240.0
vc1dsp.vc1_inv_trans_8x4_rvv_i32: 239.2   235.2

> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=78e1565f847588cb3d34d1a920902fb253d9c03c
---

 libavcodec/riscv/vc1dsp_rvv.S | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/libavcodec/riscv/vc1dsp_rvv.S b/libavcodec/riscv/vc1dsp_rvv.S
index 9d85377cec..8c127c7644 100644
--- a/libavcodec/riscv/vc1dsp_rvv.S
+++ b/libavcodec/riscv/vc1dsp_rvv.S
@@ -194,14 +194,12 @@ func ff_vc1_inv_trans_4_rvv, zve32x
         li      t4, 22
         vmul.vx v10, v2, t3
         li      t2, 10
-        vmul.vx v14, v1, t4
+        vmul.vx v26, v1, t4
+        vmul.vx v27, v3, t4
         vadd.vv v24, v8, v10  # t1
         vsub.vv v25, v8, v10  # t2
-        vmul.vx v16, v3, t2
-        vmul.vx v18, v3, t4
-        vmul.vx v20, v1, t2
-        vadd.vv v26, v14, v16 # t3
-        vsub.vv v27, v18, v20 # t4
+        vmacc.vx v26, t2, v3  # t3
+        vnmsac.vx v27, t2, v1 # t4
         vwadd.vv  v8, v24, v26
         vwsub.vv  v10, v25, v27
         vwadd.vv  v12, v25, v27