[FFmpeg-devel] Subject: [PATCH 3/3] lavc/dnxhdenc: R-V V get_pixels_8x4_sym

[Rémi Denis-Courmont]

Hi,

+/*
+ * Copyright (c) 2023 Institue of Software Chinese Academy of Sciences 
(ISCAS).
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 
USA
+ */
+
+#include "libavutil/riscv/asm.S"
+
+func ff_get_pixels_8x4_sym_rvv, zve64x
+        vsetivli    zero, 8, e8, mf2, ta, ma
+        vlse64.v    v16, (a1), a2
+        li          t0, 8 * 8
+        vsetvli     zero, t0, e16, m4, ta, ma
+        vzext.vf2   v8, v16
+        vse16.v     v8, (a0)
+        li          a2, 8*2

That's not needed. You can use immediate values.

+        vsetivli    zero, 2, e8, mf8, ta, ma
+        addi        a1, a0, 48
+        addi        a0, a0, 32*2
+        vle64.v     v0, (a1)
+        vse64.v     v0, (a0)
+        sub         a1, a1, a2
+        vle64.v     v0, (a1)
+        add         a0, a0, a2
+        vse64.v     v0, (a0)
+        sub         a1, a1, a2
+        vle64.v     v0, (a1)
+        add         a0, a0, a2
+        vse64.v     v0, (a0)
+        sub         a1, a1, a2
+        vle64.v     v0, (a1)
+        add         a0, a0, a2
+        vse64.v     v0, (a0)

You can reorder to avoid immediate data dependencies on the addresses.

I expect that it would be faster to make one large load, and then 4 small 
stores, but that might work only for exactly 128-bit vectors?

In any case, you need to check the vector length in init.

+
+        ret
+endfunc

-- 
雷米‧德尼-库尔蒙
http://www.remlab.net/