aarch64/vvc: Add sad

sad_8x16_c:                                              0.8 ( 1.00x)
sad_8x16_neon:                                           0.2 ( 3.00x)
sad_16x8_c:                                              0.5 ( 1.00x)
sad_16x8_neon:                                           0.2 ( 2.00x)
sad_16x16_c:                                             1.5 ( 1.00x)
sad_16x16_neon:                                          0.2 ( 6.00x)

libavcodec/aarch64/vvc/Makefile

@@ -3,6 +3,7 @@ clean::
 
 OBJS-$(CONFIG_VVC_DECODER)              += aarch64/vvc/dsp_init.o
 NEON-OBJS-$(CONFIG_VVC_DECODER)         += aarch64/vvc/alf.o \
+                                           aarch64/vvc/sad.o \
                                            aarch64/h26x/epel_neon.o \
                                            aarch64/h26x/qpel_neon.o \
                                            aarch64/h26x/sao_neon.o

libavcodec/aarch64/vvc/dsp_init.c

@@ -39,6 +39,9 @@
 #include "alf_template.c"
 #undef BIT_DEPTH
 
+int ff_vvc_sad_neon(const int16_t *src0, const int16_t *src1, int dx, int dy,
+                    const int block_w, const int block_h);
+
 void ff_vvc_dsp_init_aarch64(VVCDSPContext *const c, const int bd)
 {
     int cpu_flags = av_get_cpu_flags();
@@ -125,4 +128,6 @@ void ff_vvc_dsp_init_aarch64(VVCDSPContext *const c, const int bd)
         c->alf.filter[LUMA] = alf_filter_luma_12_neon;
         c->alf.filter[CHROMA] = alf_filter_chroma_12_neon;
     }
+
+    c->inter.sad = ff_vvc_sad_neon;
 }

libavcodec/aarch64/vvc/sad.S

@@ -0,0 +1,75 @@
+/*
+ * Copyright (c) 2024 Zhao Zhili <quinkblack@foxmail.com>
+ *
+ * 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/aarch64/asm.S"
+
+#define VVC_MAX_PB_SIZE 128
+
+function ff_vvc_sad_neon, export=1
+        src0            .req x0
+        src1            .req x1
+        dx              .req w2
+        dy              .req w3
+        block_w         .req w4
+        block_h         .req w5
+
+        sub             w7, dx, #4
+        sub             w8, dy, #4
+        add             w6, dx, dy, lsl #7
+        add             w7, w7, w8, lsl #7
+        sxtw            x6, w6
+        sxtw            x7, w7
+        add             src0, src0, x6, lsl #1
+        sub             src1, src1, x7, lsl #1
+
+        cmp             block_w, #16
+        movi            v16.4s, #0
+        b.ge            2f
+1:
+        // block_w == 8
+        ldr             q0, [src0]
+        ldr             q2, [src1]
+        subs            block_h, block_h, #2
+        sabal           v16.4s, v0.4h, v2.4h
+        sabal2          v16.4s, v0.8h, v2.8h
+
+        add             src0, src0, #(2 * VVC_MAX_PB_SIZE * 2)
+        add             src1, src1, #(2 * VVC_MAX_PB_SIZE * 2)
+        b.ne            1b
+        b               4f
+2:
+        // block_w == 16, no block_w > 16 according the spec
+        movi            v17.4s, #0
+3:
+        ldp             q0, q1, [src0], #(2 * VVC_MAX_PB_SIZE * 2)
+        ldp             q2, q3, [src1], #(2 * VVC_MAX_PB_SIZE * 2)
+        subs            block_h, block_h, #2
+        sabal           v16.4s, v0.4h, v2.4h
+        sabal2          v16.4s, v0.8h, v2.8h
+        sabal           v17.4s, v1.4h, v3.4h
+        sabal2          v17.4s, v1.8h, v3.8h
+
+        b.ne            3b
+        add             v16.4s, v16.4s, v17.4s
+4:
+        addv            s16, v16.4s
+        mov             w0, v16.s[0]
+        ret
+endfunc