@@ -40,6 +40,7 @@ AVCODECOBJS-$(CONFIG_V210_DECODER) += v210dec.o
AVCODECOBJS-$(CONFIG_V210_ENCODER) += v210enc.o
AVCODECOBJS-$(CONFIG_VORBIS_DECODER) += vorbisdsp.o
AVCODECOBJS-$(CONFIG_VP9_DECODER) += vp9dsp.o
+AVCODECOBJS-$(CONFIG_VVC_DECODER) += vvc_mc.o
CHECKASMOBJS-$(CONFIG_AVCODEC) += $(AVCODECOBJS-yes)
@@ -194,6 +194,9 @@ static const struct {
#if CONFIG_VORBIS_DECODER
{ "vorbisdsp", checkasm_check_vorbisdsp },
#endif
+ #if CONFIG_VVC_DECODER
+ { "vvc_mc", checkasm_check_vvc_mc },
+ #endif
#endif
#if CONFIG_AVFILTER
#if CONFIG_AFIR_FILTER
@@ -131,6 +131,7 @@ void checkasm_check_vp8dsp(void);
void checkasm_check_vp9dsp(void);
void checkasm_check_videodsp(void);
void checkasm_check_vorbisdsp(void);
+void checkasm_check_vvc_mc(void);
struct CheckasmPerf;
new file mode 100644
@@ -0,0 +1,270 @@
+/*
+ * Copyright (c) 2023-2024 Nuo Mi
+ * Copyright (c) 2023-2024 Wu Jianhua
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 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 General Public License for more details.
+ *
+ * You should have received a copy of the GNU 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 <string.h>
+
+#include "checkasm.h"
+#include "libavcodec/avcodec.h"
+#include "libavcodec/vvc/vvc_ctu.h"
+#include "libavcodec/vvc/vvc_data.h"
+
+#include "libavutil/common.h"
+#include "libavutil/internal.h"
+#include "libavutil/internal.h"
+#include "libavutil/intreadwrite.h"
+#include "libavutil/mem_internal.h"
+
+static const uint32_t pixel_mask[] = { 0xffffffff, 0x03ff03ff, 0x0fff0fff, 0x3fff3fff, 0xffffffff };
+static const int sizes[] = { 2, 4, 8, 16, 32, 64, 128 };
+
+#define PIXEL_STRIDE (MAX_CTU_SIZE * 2)
+#define EXTRA_BEFORE 3
+#define EXTRA_AFTER 4
+#define SRC_EXTRA (EXTRA_BEFORE + EXTRA_AFTER) * 2
+#define SRC_BUF_SIZE (PIXEL_STRIDE + SRC_EXTRA) * (PIXEL_STRIDE + SRC_EXTRA)
+#define DST_BUF_SIZE (MAX_CTU_SIZE * MAX_CTU_SIZE * 2)
+#define SRC_OFFSET ((PIXEL_STRIDE + EXTRA_BEFORE * 2) * EXTRA_BEFORE)
+
+#define randomize_buffers(buf0, buf1, size, mask) \
+ do { \
+ int k; \
+ for (k = 0; k < size; k += 4) { \
+ uint32_t r = rnd() & mask; \
+ AV_WN32A(buf0 + k, r); \
+ AV_WN32A(buf1 + k, r); \
+ } \
+ } while (0)
+
+#define randomize_pixels(buf0, buf1, size) \
+ do { \
+ uint32_t mask = pixel_mask[(bit_depth - 8) >> 1]; \
+ randomize_buffers(buf0, buf1, size, mask); \
+ } while (0)
+
+#define randomize_avg_src(buf0, buf1, size) \
+ do { \
+ uint32_t mask = 0x3fff3fff; \
+ randomize_buffers(buf0, buf1, size, mask); \
+ } while (0)
+
+static void check_put_vvc_luma(void)
+{
+ LOCAL_ALIGNED_32(int16_t, dst0, [DST_BUF_SIZE / 2]);
+ LOCAL_ALIGNED_32(int16_t, dst1, [DST_BUF_SIZE / 2]);
+ LOCAL_ALIGNED_32(uint8_t, src0, [SRC_BUF_SIZE]);
+ LOCAL_ALIGNED_32(uint8_t, src1, [SRC_BUF_SIZE]);
+ VVCDSPContext c;
+
+ declare_func_emms(AV_CPU_FLAG_MMX | AV_CPU_FLAG_MMXEXT, void, int16_t *dst, const uint8_t *src, const ptrdiff_t src_stride,
+ const int height, const int8_t *hf, const int8_t *vf, const int width);
+
+ for (int bit_depth = 8; bit_depth <= 12; bit_depth += 2) {
+ randomize_pixels(src0, src1, SRC_BUF_SIZE);
+ ff_vvc_dsp_init(&c, bit_depth);
+ for (int i = 0; i < 2; i++) {
+ for (int j = 0; j < 2; j++) {
+ for (int h = 4; h <= MAX_CTU_SIZE; h *= 2) {
+ for (int w = 4; w <= MAX_CTU_SIZE; w *= 2) {
+ const int idx = av_log2(w) - 1;
+ const int mx = rnd() % 16;
+ const int my = rnd() % 16;
+ const int8_t *hf = ff_vvc_inter_luma_filters[rnd() % 3][mx];
+ const int8_t *vf = ff_vvc_inter_luma_filters[rnd() % 3][my];
+ const char *type;
+ switch ((j << 1) | i) {
+ case 0: type = "put_luma_pixels"; break; // 0 0
+ case 1: type = "put_luma_h"; break; // 0 1
+ case 2: type = "put_luma_v"; break; // 1 0
+ case 3: type = "put_luma_hv"; break; // 1 1
+ }
+ if (check_func(c.inter.put[LUMA][idx][j][i], "%s_%d_%dx%d", type, bit_depth, w, h)) {
+ memset(dst0, 0, DST_BUF_SIZE);
+ memset(dst1, 0, DST_BUF_SIZE);
+ call_ref(dst0, src0 + SRC_OFFSET, PIXEL_STRIDE, h, hf, vf, w);
+ call_new(dst1, src1 + SRC_OFFSET, PIXEL_STRIDE, h, hf, vf, w);
+ if (memcmp(dst0, dst1, DST_BUF_SIZE))
+ fail();
+ if (w == h)
+ bench_new(dst1, src1 + SRC_OFFSET, PIXEL_STRIDE, h, hf, vf, w);
+ }
+ }
+ }
+ }
+ }
+ }
+ report("put_luma");
+}
+
+static void check_put_vvc_luma_uni(void)
+{
+ LOCAL_ALIGNED_32(uint8_t, dst0, [DST_BUF_SIZE]);
+ LOCAL_ALIGNED_32(uint8_t, dst1, [DST_BUF_SIZE]);
+ LOCAL_ALIGNED_32(uint8_t, src0, [SRC_BUF_SIZE]);
+ LOCAL_ALIGNED_32(uint8_t, src1, [SRC_BUF_SIZE]);
+
+ VVCDSPContext c;
+ declare_func_emms(AV_CPU_FLAG_MMX | AV_CPU_FLAG_MMXEXT, void, uint8_t *dst, ptrdiff_t dststride,
+ uint8_t *src, ptrdiff_t srcstride, int height, const int8_t *hf, const int8_t *vf, int width);
+
+ for (int bit_depth = 8; bit_depth <= 12; bit_depth += 2) {
+ ff_vvc_dsp_init(&c, bit_depth);
+ randomize_pixels(src0, src1, SRC_BUF_SIZE);
+ for (int i = 0; i < 2; i++) {
+ for (int j = 0; j < 2; j++) {
+ for (int h = 4; h <= MAX_CTU_SIZE; h *= 2) {
+ for (int w = 4; w <= MAX_CTU_SIZE; w *= 2) {
+ const int idx = av_log2(w) - 1;
+ const int mx = rnd() % VVC_INTER_LUMA_FACTS;
+ const int my = rnd() % VVC_INTER_LUMA_FACTS;
+ const int8_t *hf = ff_vvc_inter_luma_filters[rnd() % VVC_INTER_FILTER_TYPES][mx];
+ const int8_t *vf = ff_vvc_inter_luma_filters[rnd() % VVC_INTER_FILTER_TYPES][my];
+ const char *type;
+
+ switch ((j << 1) | i) {
+ case 0: type = "put_uni_pixels"; break; // 0 0
+ case 1: type = "put_uni_h"; break; // 0 1
+ case 2: type = "put_uni_v"; break; // 1 0
+ case 3: type = "put_uni_hv"; break; // 1 1
+ }
+
+ if (check_func(c.inter.put_uni[LUMA][idx][j][i], "%s_luma_%d_%dx%d", type, bit_depth, w, h)) {
+ memset(dst0, 0, DST_BUF_SIZE);
+ memset(dst1, 0, DST_BUF_SIZE);
+ call_ref(dst0, PIXEL_STRIDE, src0 + SRC_OFFSET, PIXEL_STRIDE, h, hf, vf, w);
+ call_new(dst1, PIXEL_STRIDE, src1 + SRC_OFFSET, PIXEL_STRIDE, h, hf, vf, w);
+ if (memcmp(dst0, dst1, DST_BUF_SIZE))
+ fail();
+ if (w == h)
+ bench_new(dst1, PIXEL_STRIDE, src1 + SRC_OFFSET, PIXEL_STRIDE, h, hf, vf, w);
+ }
+ }
+ }
+ }
+ }
+ }
+ report("put_uni_luma");
+}
+
+static void check_put_vvc_chroma(void)
+{
+ LOCAL_ALIGNED_32(int16_t, dst0, [DST_BUF_SIZE / 2]);
+ LOCAL_ALIGNED_32(int16_t, dst1, [DST_BUF_SIZE / 2]);
+ LOCAL_ALIGNED_32(uint8_t, src0, [SRC_BUF_SIZE]);
+ LOCAL_ALIGNED_32(uint8_t, src1, [SRC_BUF_SIZE]);
+ VVCDSPContext c;
+
+ declare_func_emms(AV_CPU_FLAG_MMX | AV_CPU_FLAG_MMXEXT, void, int16_t *dst, const uint8_t *src, const ptrdiff_t src_stride,
+ const int height, const int8_t *hf, const int8_t *vf, const int width);
+
+ for (int bit_depth = 8; bit_depth <= 12; bit_depth += 2) {
+ randomize_pixels(src0, src1, SRC_BUF_SIZE);
+ ff_vvc_dsp_init(&c, bit_depth);
+ for (int i = 0; i < 2; i++) {
+ for (int j = 0; j < 2; j++) {
+ for (int h = 2; h <= MAX_CTU_SIZE; h *= 2) {
+ for (int w = 2; w <= MAX_CTU_SIZE; w *= 2) {
+ const int idx = av_log2(w) - 1;
+ const int mx = rnd() % VVC_INTER_CHROMA_FACTS;
+ const int my = rnd() % VVC_INTER_CHROMA_FACTS;
+ const int8_t *hf = ff_vvc_inter_chroma_filters[rnd() % VVC_INTER_FILTER_TYPES][mx];
+ const int8_t *vf = ff_vvc_inter_chroma_filters[rnd() % VVC_INTER_FILTER_TYPES][my];
+ const char *type;
+ switch ((j << 1) | i) {
+ case 0: type = "put_chroma_pixels"; break; // 0 0
+ case 1: type = "put_chroma_h"; break; // 0 1
+ case 2: type = "put_chroma_v"; break; // 1 0
+ case 3: type = "put_chroma_hv"; break; // 1 1
+ }
+ if (check_func(c.inter.put[CHROMA][idx][j][i], "%s_%d_%dx%d", type, bit_depth, w, h)) {
+ memset(dst0, 0, DST_BUF_SIZE);
+ memset(dst1, 0, DST_BUF_SIZE);
+ call_ref(dst0, src0 + SRC_OFFSET, PIXEL_STRIDE, h, hf, vf, w);
+ call_new(dst1, src1 + SRC_OFFSET, PIXEL_STRIDE, h, hf, vf, w);
+ if (memcmp(dst0, dst1, DST_BUF_SIZE))
+ fail();
+ if (w == h)
+ bench_new(dst1, src1 + SRC_OFFSET, PIXEL_STRIDE, h, hf, vf, w);
+ }
+ }
+ }
+ }
+ }
+ }
+ report("put_chroma");
+}
+
+static void check_put_vvc_chroma_uni(void)
+{
+ LOCAL_ALIGNED_32(uint8_t, dst0, [DST_BUF_SIZE]);
+ LOCAL_ALIGNED_32(uint8_t, dst1, [DST_BUF_SIZE]);
+ LOCAL_ALIGNED_32(uint8_t, src0, [SRC_BUF_SIZE]);
+ LOCAL_ALIGNED_32(uint8_t, src1, [SRC_BUF_SIZE]);
+
+ VVCDSPContext c;
+ declare_func_emms(AV_CPU_FLAG_MMX | AV_CPU_FLAG_MMXEXT, void, uint8_t *dst, ptrdiff_t dststride,
+ uint8_t *src, ptrdiff_t srcstride, int height, const int8_t *hf, const int8_t *vf, int width);
+
+ for (int bit_depth = 8; bit_depth <= 12; bit_depth += 2) {
+ ff_vvc_dsp_init(&c, bit_depth);
+ randomize_pixels(src0, src1, SRC_BUF_SIZE);
+ for (int i = 0; i < 2; i++) {
+ for (int j = 0; j < 2; j++) {
+ for (int h = 4; h <= MAX_CTU_SIZE; h *= 2) {
+ for (int w = 4; w <= MAX_CTU_SIZE; w *= 2) {
+ const int idx = av_log2(w) - 1;
+ const int mx = rnd() % VVC_INTER_CHROMA_FACTS;
+ const int my = rnd() % VVC_INTER_CHROMA_FACTS;
+ const int8_t *hf = ff_vvc_inter_chroma_filters[rnd() % VVC_INTER_FILTER_TYPES][mx];
+ const int8_t *vf = ff_vvc_inter_chroma_filters[rnd() % VVC_INTER_FILTER_TYPES][my];
+ const char *type;
+
+ switch ((j << 1) | i) {
+ case 0: type = "put_uni_pixels"; break; // 0 0
+ case 1: type = "put_uni_h"; break; // 0 1
+ case 2: type = "put_uni_v"; break; // 1 0
+ case 3: type = "put_uni_hv"; break; // 1 1
+ }
+
+ if (check_func(c.inter.put_uni[CHROMA][idx][j][i], "%s_chroma_%d_%dx%d", type, bit_depth, w, h)) {
+ memset(dst0, 0, DST_BUF_SIZE);
+ memset(dst1, 0, DST_BUF_SIZE);
+ call_ref(dst0, PIXEL_STRIDE, src0 + SRC_OFFSET, PIXEL_STRIDE, h, hf, vf, w);
+ call_new(dst1, PIXEL_STRIDE, src1 + SRC_OFFSET, PIXEL_STRIDE, h, hf, vf, w);
+ if (memcmp(dst0, dst1, DST_BUF_SIZE))
+ fail();
+ if (w == h)
+ bench_new(dst1, PIXEL_STRIDE, src1 + SRC_OFFSET, PIXEL_STRIDE, h, hf, vf, w);
+ }
+ }
+ }
+ }
+ }
+ }
+ report("put_uni_chroma");
+}
+
+void checkasm_check_vvc_mc(void)
+{
+ check_put_vvc_luma();
+ check_put_vvc_luma_uni();
+ check_put_vvc_chroma();
+ check_put_vvc_chroma_uni();
+}