Merge "Unify set_contexts() function for encoder and decoder" into nextgenv2
diff --git a/aom_dsp/arm/aom_convolve8_avg_neon.c b/aom_dsp/arm/aom_convolve8_avg_neon.c
index 7dc936d..09429d6 100644
--- a/aom_dsp/arm/aom_convolve8_avg_neon.c
+++ b/aom_dsp/arm/aom_convolve8_avg_neon.c
@@ -65,6 +65,10 @@
assert(x_step_q4 == 16);
+ (void)x_step_q4;
+ (void)y_step_q4;
+ (void)filter_y;
+
q0s16 = vld1q_s16(filter_x);
src -= 3; // adjust for taps
@@ -241,6 +245,10 @@
assert(y_step_q4 == 16);
+ (void)x_step_q4;
+ (void)y_step_q4;
+ (void)filter_x;
+
src -= src_stride * 3;
q0s16 = vld1q_s16(filter_y);
for (; w > 0; w -= 4, src += 4, dst += 4) { // loop_vert_h
diff --git a/aom_dsp/arm/aom_convolve8_neon.c b/aom_dsp/arm/aom_convolve8_neon.c
index ed0df6d..8ebffb5 100644
--- a/aom_dsp/arm/aom_convolve8_neon.c
+++ b/aom_dsp/arm/aom_convolve8_neon.c
@@ -65,6 +65,10 @@
assert(x_step_q4 == 16);
+ (void)x_step_q4;
+ (void)y_step_q4;
+ (void)filter_y;
+
q0s16 = vld1q_s16(filter_x);
src -= 3; // adjust for taps
@@ -225,6 +229,10 @@
assert(y_step_q4 == 16);
+ (void)x_step_q4;
+ (void)y_step_q4;
+ (void)filter_x;
+
src -= src_stride * 3;
q0s16 = vld1q_s16(filter_y);
for (; w > 0; w -= 4, src += 4, dst += 4) { // loop_vert_h
diff --git a/aom_dsp/x86/convolve.h b/aom_dsp/x86/convolve.h
index ffaed02..87ff34b 100644
--- a/aom_dsp/x86/convolve.h
+++ b/aom_dsp/x86/convolve.h
@@ -27,6 +27,10 @@
const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, \
ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, \
const int16_t *filter_y, int y_step_q4, int w, int h) { \
+ (void)filter_x; \
+ (void)x_step_q4; \
+ (void)filter_y; \
+ (void)y_step_q4; \
assert(filter[3] != 128); \
assert(step_q4 == 16); \
if (filter[0] | filter[1] | filter[2]) { \
diff --git a/aom_scale/generic/aom_scale.c b/aom_scale/generic/aom_scale.c
index 49084db..28604ac 100644
--- a/aom_scale/generic/aom_scale.c
+++ b/aom_scale/generic/aom_scale.c
@@ -68,24 +68,25 @@
unsigned int source_scale, unsigned int source_length,
unsigned char *dest, int dest_step,
unsigned int dest_scale, unsigned int dest_length) {
- unsigned int i, j;
- unsigned int temp;
- int source_pitch = source_step;
+ const unsigned int source_pitch = source_step;
+ const unsigned char *const dest_end = dest + dest_length * dest_step;
(void)source_length;
(void)source_scale;
(void)dest_scale;
- source_step *= 2;
- dest[0] = source[0];
+ source_step *= 2; // Every other row.
- for (i = dest_step, j = source_step; i < dest_length * dest_step;
- i += dest_step, j += source_step) {
- temp = 8;
- temp += 3 * source[j - source_pitch];
- temp += 10 * source[j];
- temp += 3 * source[j + source_pitch];
- temp >>= 4;
- dest[i] = (char)(temp);
+ dest[0] = source[0]; // Special case: 1st pixel.
+ source += source_step;
+ dest += dest_step;
+
+ while (dest < dest_end) {
+ const unsigned int a = 3 * source[-source_pitch];
+ const unsigned int b = 10 * source[0];
+ const unsigned int c = 3 * source[source_pitch];
+ *dest = (unsigned char)((8 + a + b + c) >> 4);
+ source += source_step;
+ dest += dest_step;
}
}
@@ -119,17 +120,18 @@
unsigned int source_length, unsigned char *dest,
int dest_step, unsigned int dest_scale,
unsigned int dest_length) {
- unsigned int i, j;
-
+ const unsigned char *const dest_end = dest + dest_length * dest_step;
(void)source_length;
(void)source_scale;
(void)dest_scale;
- source_step *= 2;
- j = 0;
+ source_step *= 2; // Every other row.
- for (i = 0; i < dest_length * dest_step; i += dest_step, j += source_step)
- dest[i] = source[j];
+ while (dest < dest_end) {
+ *dest = *source;
+ source += source_step;
+ dest += dest_step;
+ }
}
/****************************************************************************
*
@@ -159,12 +161,12 @@
unsigned int source_scale, unsigned int source_length,
unsigned char *dest, int dest_step,
unsigned int dest_scale, unsigned int dest_length) {
- unsigned int i;
- unsigned int round_value = dest_scale / 2;
+ const unsigned char *const dest_end = dest + dest_length * dest_step;
+ const unsigned int round_value = dest_scale / 2;
unsigned int left_modifier = dest_scale;
unsigned int right_modifier = 0;
- unsigned char left_pixel = *source;
- unsigned char right_pixel = *(source + source_step);
+ unsigned char left_pixel = source[0];
+ unsigned char right_pixel = source[source_step];
(void)source_length;
@@ -173,18 +175,18 @@
/* assert ( (source_length - 1) * dest_scale >= (dest_length - 1) *
* source_scale);*/
- for (i = 0; i < dest_length * dest_step; i += dest_step) {
- dest[i] = (char)((left_modifier * left_pixel +
- right_modifier * right_pixel + round_value) /
- dest_scale);
+ while (dest < dest_end) {
+ *dest = (unsigned char)((left_modifier * left_pixel +
+ right_modifier * right_pixel + round_value) /
+ dest_scale);
right_modifier += source_scale;
while (right_modifier > dest_scale) {
right_modifier -= dest_scale;
source += source_step;
- left_pixel = *source;
- right_pixel = *(source + source_step);
+ left_pixel = source[0];
+ right_pixel = source[source_step];
}
left_modifier = dest_scale - right_modifier;
@@ -236,11 +238,10 @@
unsigned int dest_width, unsigned int dest_height, unsigned char *temp_area,
unsigned char temp_area_height, unsigned int hscale, unsigned int hratio,
unsigned int vscale, unsigned int vratio, unsigned int interlaced) {
- /*unsigned*/
- int i, j, k;
- int bands;
- int dest_band_height;
- int source_band_height;
+ unsigned int i, j, k;
+ unsigned int bands;
+ unsigned int dest_band_height;
+ unsigned int source_band_height;
typedef void (*Scale1D)(const unsigned char *source, int source_step,
unsigned int source_scale, unsigned int source_length,
@@ -331,7 +332,7 @@
if (ratio_scalable) {
if (source_height == dest_height) {
/* for each band of the image */
- for (k = 0; k < (int)dest_height; k++) {
+ for (k = 0; k < dest_height; ++k) {
horiz_line_scale(source, source_width, dest, dest_width);
source += source_pitch;
dest += dest_pitch;
@@ -346,14 +347,13 @@
horiz_line_scale(source, source_width, temp_area, dest_width);
}
- for (k = 0;
- k < (int)(dest_height + dest_band_height - 1) / dest_band_height;
- k++) {
+ for (k = 0; k < (dest_height + dest_band_height - 1) / dest_band_height;
+ ++k) {
/* scale one band horizontally */
- for (i = 0; i < source_band_height; i++) {
+ for (i = 0; i < source_band_height; ++i) {
/* Trap case where we could read off the base of the source buffer */
- line_src = (unsigned char *)source + i * source_pitch;
+ line_src = source + i * source_pitch;
if (line_src < source_base) line_src = source_base;
@@ -388,7 +388,7 @@
if (source_height == dest_height) {
/* for each band of the image */
- for (k = 0; k < (int)dest_height; k++) {
+ for (k = 0; k < dest_height; ++k) {
Scale1Dh(source, 1, hscale, source_width + 1, dest, 1, hratio,
dest_width);
source += source_pitch;
@@ -414,10 +414,10 @@
/* for each band of the image */
bands = (dest_height + dest_band_height - 1) / dest_band_height;
- for (k = 0; k < bands; k++) {
+ for (k = 0; k < bands; ++k) {
/* scale one band horizontally */
- for (i = 1; i < source_band_height + 1; i++) {
- if (k * source_band_height + i < (int)source_height) {
+ for (i = 1; i < source_band_height + 1; ++i) {
+ if (k * source_band_height + i < source_height) {
Scale1Dh(source + i * source_pitch, 1, hscale, source_width + 1,
temp_area + i * dest_pitch, 1, hratio, dest_width);
} else { /* Duplicate the last row */
@@ -428,7 +428,7 @@
}
/* scale one band vertically */
- for (j = 0; j < (int)dest_width; j++) {
+ for (j = 0; j < dest_width; ++j) {
Scale1Dv(&temp_area[j], dest_pitch, vscale, source_band_height + 1,
&dest[j], dest_pitch, vratio, dest_band_height);
}
@@ -487,12 +487,12 @@
temp_area, temp_height, hscale, hratio, vscale, vratio, interlaced);
if (dw < (int)dst->y_width)
- for (i = 0; i < dh; i++)
+ for (i = 0; i < dh; ++i)
memset(dst->y_buffer + i * dst->y_stride + dw - 1,
dst->y_buffer[i * dst->y_stride + dw - 2], dst->y_width - dw + 1);
if (dh < (int)dst->y_height)
- for (i = dh - 1; i < (int)dst->y_height; i++)
+ for (i = dh - 1; i < (int)dst->y_height; ++i)
memcpy(dst->y_buffer + i * dst->y_stride,
dst->y_buffer + (dh - 2) * dst->y_stride, dst->y_width + 1);
@@ -502,13 +502,13 @@
vratio, interlaced);
if (dw / 2 < (int)dst->uv_width)
- for (i = 0; i < dst->uv_height; i++)
+ for (i = 0; i < dst->uv_height; ++i)
memset(dst->u_buffer + i * dst->uv_stride + dw / 2 - 1,
dst->u_buffer[i * dst->uv_stride + dw / 2 - 2],
dst->uv_width - dw / 2 + 1);
if (dh / 2 < (int)dst->uv_height)
- for (i = dh / 2 - 1; i < (int)dst->y_height / 2; i++)
+ for (i = dh / 2 - 1; i < (int)dst->y_height / 2; ++i)
memcpy(dst->u_buffer + i * dst->uv_stride,
dst->u_buffer + (dh / 2 - 2) * dst->uv_stride, dst->uv_width);
@@ -518,13 +518,13 @@
vratio, interlaced);
if (dw / 2 < (int)dst->uv_width)
- for (i = 0; i < dst->uv_height; i++)
+ for (i = 0; i < dst->uv_height; ++i)
memset(dst->v_buffer + i * dst->uv_stride + dw / 2 - 1,
dst->v_buffer[i * dst->uv_stride + dw / 2 - 2],
dst->uv_width - dw / 2 + 1);
if (dh / 2 < (int)dst->uv_height)
- for (i = dh / 2 - 1; i < (int)dst->y_height / 2; i++)
+ for (i = dh / 2 - 1; i < (int)dst->y_height / 2; ++i)
memcpy(dst->v_buffer + i * dst->uv_stride,
dst->v_buffer + (dh / 2 - 2) * dst->uv_stride, dst->uv_width);
}
diff --git a/aom_scale/generic/gen_scalers.c b/aom_scale/generic/gen_scalers.c
index 57c464d..fd638bd 100644
--- a/aom_scale/generic/gen_scalers.c
+++ b/aom_scale/generic/gen_scalers.c
@@ -39,27 +39,23 @@
unsigned int source_width,
unsigned char *dest,
unsigned int dest_width) {
- unsigned i;
- unsigned int a, b, c, d, e;
- unsigned char *des = dest;
- const unsigned char *src = source;
-
+ const unsigned char *const source_end = source + source_width;
(void)dest_width;
- for (i = 0; i < source_width; i += 5) {
- a = src[0];
- b = src[1];
- c = src[2];
- d = src[3];
- e = src[4];
+ while (source < source_end) {
+ const unsigned int a = source[0];
+ const unsigned int b = source[1];
+ const unsigned int c = source[2];
+ const unsigned int d = source[3];
+ const unsigned int e = source[4];
- des[0] = (unsigned char)a;
- des[1] = (unsigned char)((b * 192 + c * 64 + 128) >> 8);
- des[2] = (unsigned char)((c * 128 + d * 128 + 128) >> 8);
- des[3] = (unsigned char)((d * 64 + e * 192 + 128) >> 8);
+ dest[0] = (unsigned char)a;
+ dest[1] = (unsigned char)((b * 192 + c * 64 + 128) >> 8);
+ dest[2] = (unsigned char)((c * 128 + d * 128 + 128) >> 8);
+ dest[3] = (unsigned char)((d * 64 + e * 192 + 128) >> 8);
- src += 5;
- des += 4;
+ source += 5;
+ dest += 4;
}
}
@@ -67,25 +63,21 @@
unsigned int src_pitch, unsigned char *dest,
unsigned int dest_pitch,
unsigned int dest_width) {
- unsigned int i;
- unsigned int a, b, c, d, e;
- unsigned char *des = dest;
- unsigned char *src = source;
+ const unsigned char *const dest_end = dest + dest_width;
+ while (dest < dest_end) {
+ const unsigned int a = source[0 * src_pitch];
+ const unsigned int b = source[1 * src_pitch];
+ const unsigned int c = source[2 * src_pitch];
+ const unsigned int d = source[3 * src_pitch];
+ const unsigned int e = source[4 * src_pitch];
- for (i = 0; i < dest_width; i++) {
- a = src[0 * src_pitch];
- b = src[1 * src_pitch];
- c = src[2 * src_pitch];
- d = src[3 * src_pitch];
- e = src[4 * src_pitch];
+ dest[0 * dest_pitch] = (unsigned char)a;
+ dest[1 * dest_pitch] = (unsigned char)((b * 192 + c * 64 + 128) >> 8);
+ dest[2 * dest_pitch] = (unsigned char)((c * 128 + d * 128 + 128) >> 8);
+ dest[3 * dest_pitch] = (unsigned char)((d * 64 + e * 192 + 128) >> 8);
- des[0 * dest_pitch] = (unsigned char)a;
- des[1 * dest_pitch] = (unsigned char)((b * 192 + c * 64 + 128) >> 8);
- des[2 * dest_pitch] = (unsigned char)((c * 128 + d * 128 + 128) >> 8);
- des[3 * dest_pitch] = (unsigned char)((d * 64 + e * 192 + 128) >> 8);
-
- src++;
- des++;
+ ++source;
+ ++dest;
}
}
@@ -114,26 +106,21 @@
unsigned int source_width,
unsigned char *dest,
unsigned int dest_width) {
- unsigned int i;
- unsigned int a, b, c, d, e;
- unsigned char *des = dest;
- const unsigned char *src = source;
-
+ const unsigned char *const source_end = source + source_width;
(void)dest_width;
+ while (source < source_end) {
+ const unsigned int a = source[0];
+ const unsigned int b = source[1];
+ const unsigned int c = source[2];
+ const unsigned int d = source[3];
+ const unsigned int e = source[4];
- for (i = 0; i < source_width; i += 5) {
- a = src[0];
- b = src[1];
- c = src[2];
- d = src[3];
- e = src[4];
+ dest[0] = (unsigned char)a;
+ dest[1] = (unsigned char)((b * 85 + c * 171 + 128) >> 8);
+ dest[2] = (unsigned char)((d * 171 + e * 85 + 128) >> 8);
- des[0] = (unsigned char)a;
- des[1] = (unsigned char)((b * 85 + c * 171 + 128) >> 8);
- des[2] = (unsigned char)((d * 171 + e * 85 + 128) >> 8);
-
- src += 5;
- des += 3;
+ source += 5;
+ dest += 3;
}
}
@@ -141,24 +128,20 @@
unsigned int src_pitch, unsigned char *dest,
unsigned int dest_pitch,
unsigned int dest_width) {
- unsigned int i;
- unsigned int a, b, c, d, e;
- unsigned char *des = dest;
- unsigned char *src = source;
+ const unsigned char *const dest_end = dest + dest_width;
+ while (dest < dest_end) {
+ const unsigned int a = source[0 * src_pitch];
+ const unsigned int b = source[1 * src_pitch];
+ const unsigned int c = source[2 * src_pitch];
+ const unsigned int d = source[3 * src_pitch];
+ const unsigned int e = source[4 * src_pitch];
- for (i = 0; i < dest_width; i++) {
- a = src[0 * src_pitch];
- b = src[1 * src_pitch];
- c = src[2 * src_pitch];
- d = src[3 * src_pitch];
- e = src[4 * src_pitch];
+ dest[0 * dest_pitch] = (unsigned char)a;
+ dest[1 * dest_pitch] = (unsigned char)((b * 85 + c * 171 + 128) >> 8);
+ dest[2 * dest_pitch] = (unsigned char)((d * 171 + e * 85 + 128) >> 8);
- des[0 * dest_pitch] = (unsigned char)a;
- des[1 * dest_pitch] = (unsigned char)((b * 85 + c * 171 + 128) >> 8);
- des[2 * dest_pitch] = (unsigned char)((d * 171 + e * 85 + 128) >> 8);
-
- src++;
- des++;
+ ++source;
+ ++dest;
}
}
@@ -186,18 +169,12 @@
unsigned int source_width,
unsigned char *dest,
unsigned int dest_width) {
- unsigned int i;
- unsigned int a;
- unsigned char *des = dest;
- const unsigned char *src = source;
-
+ const unsigned char *const source_end = source + source_width;
(void)dest_width;
-
- for (i = 0; i < source_width; i += 2) {
- a = src[0];
- des[0] = (unsigned char)(a);
- src += 2;
- des += 1;
+ while (source < source_end) {
+ dest[0] = source[0];
+ source += 2;
+ ++dest;
}
}
@@ -215,18 +192,14 @@
unsigned char *dest,
unsigned int dest_pitch,
unsigned int dest_width) {
- int i;
- int temp;
- int width = dest_width;
-
+ const unsigned char *const dest_end = dest + dest_width;
(void)dest_pitch;
-
- for (i = 0; i < width; i++) {
- temp = 8;
- temp += source[i - (int)src_pitch] * 3;
- temp += source[i] * 10;
- temp += source[i + src_pitch] * 3;
- temp >>= 4;
- dest[i] = (unsigned char)(temp);
+ while (dest < dest_end) {
+ const unsigned int a = source[-src_pitch] * 3;
+ const unsigned int b = source[0] * 10;
+ const unsigned int c = source[src_pitch] * 3;
+ dest[0] = (unsigned char)((8 + a + b + c) >> 4);
+ ++source;
+ ++dest;
}
}
diff --git a/aomenc.c b/aomenc.c
index 8eb30ed..497c8d5 100644
--- a/aomenc.c
+++ b/aomenc.c
@@ -1415,9 +1415,8 @@
#if CONFIG_WEBM_IO
if (stream->config.write_webm) {
stream->webm_ctx.stream = stream->file;
- write_webm_file_header(&stream->webm_ctx, cfg, &global->framerate,
- stream->config.stereo_fmt, global->codec->fourcc,
- pixel_aspect_ratio);
+ write_webm_file_header(&stream->webm_ctx, cfg, stream->config.stereo_fmt,
+ global->codec->fourcc, pixel_aspect_ratio);
}
#else
(void)pixel_aspect_ratio;
diff --git a/av1/av1_dx_iface.c b/av1/av1_dx_iface.c
index 7da80f0..43cc3a2 100644
--- a/av1/av1_dx_iface.c
+++ b/av1/av1_dx_iface.c
@@ -828,7 +828,7 @@
static aom_codec_err_t ctrl_copy_reference(aom_codec_alg_priv_t *ctx,
va_list args) {
- aom_ref_frame_t *data = va_arg(args, aom_ref_frame_t *);
+ const aom_ref_frame_t *const frame = va_arg(args, aom_ref_frame_t *);
// Only support this function in serial decode.
if (ctx->frame_parallel_decode) {
@@ -836,8 +836,7 @@
return AOM_CODEC_INCAPABLE;
}
- if (data) {
- aom_ref_frame_t *frame = (aom_ref_frame_t *)data;
+ if (frame) {
YV12_BUFFER_CONFIG sd;
AVxWorker *const worker = ctx->frame_workers;
FrameWorkerData *const frame_worker_data = (FrameWorkerData *)worker->data1;
diff --git a/av1/common/av1_rtcd_defs.pl b/av1/common/av1_rtcd_defs.pl
index 53c9dc5..e66826f 100644
--- a/av1/common/av1_rtcd_defs.pl
+++ b/av1/common/av1_rtcd_defs.pl
@@ -391,6 +391,9 @@
add_proto qw/void av1_fht16x16/, "const int16_t *input, tran_low_t *output, int stride, int tx_type";
specialize qw/av1_fht16x16 sse2 avx2/;
+add_proto qw/void av1_fht32x32/, "const int16_t *input, tran_low_t *output, int stride, int tx_type";
+specialize qw/av1_fht32x32 avx2/;
+
if (aom_config("CONFIG_EXT_TX") eq "yes") {
add_proto qw/void av1_fht4x8/, "const int16_t *input, tran_low_t *output, int stride, int tx_type";
specialize qw/av1_fht4x8 sse2/;
@@ -409,9 +412,6 @@
add_proto qw/void av1_fht32x16/, "const int16_t *input, tran_low_t *output, int stride, int tx_type";
specialize qw/av1_fht32x16 sse2/;
-
- add_proto qw/void av1_fht32x32/, "const int16_t *input, tran_low_t *output, int stride, int tx_type";
- specialize qw/av1_fht32x32 avx2/;
}
if (aom_config("CONFIG_EMULATE_HARDWARE") eq "yes") {
diff --git a/av1/common/entropymode.c b/av1/common/entropymode.c
index 20e8904..78f4ffe 100644
--- a/av1/common/entropymode.c
+++ b/av1/common/entropymode.c
@@ -892,13 +892,14 @@
#if CONFIG_PALETTE
int av1_get_palette_color_context(const uint8_t *color_map, int cols, int r,
- int c, int n, int *color_order) {
+ int c, int n, uint8_t *color_order,
+ int *color_idx) {
int i, j, max, max_idx, temp;
int scores[PALETTE_MAX_SIZE + 10];
int weights[4] = { 3, 2, 3, 2 };
int color_ctx = 0;
int color_neighbors[4];
-
+ int inverse_color_order[PALETTE_MAX_SIZE];
assert(n <= PALETTE_MAX_SIZE);
if (c - 1 >= 0)
@@ -918,7 +919,10 @@
else
color_neighbors[3] = -1;
- for (i = 0; i < PALETTE_MAX_SIZE; ++i) color_order[i] = i;
+ for (i = 0; i < PALETTE_MAX_SIZE; ++i) {
+ color_order[i] = i;
+ inverse_color_order[i] = i;
+ }
memset(scores, 0, PALETTE_MAX_SIZE * sizeof(scores[0]));
for (i = 0; i < 4; ++i) {
if (color_neighbors[i] >= 0) scores[color_neighbors[i]] += weights[i];
@@ -944,6 +948,8 @@
temp = color_order[i];
color_order[i] = color_order[max_idx];
color_order[max_idx] = temp;
+ inverse_color_order[color_order[i]] = i;
+ inverse_color_order[color_order[max_idx]] = max_idx;
}
}
@@ -956,7 +962,9 @@
}
if (color_ctx >= PALETTE_COLOR_CONTEXTS) color_ctx = 0;
-
+ if (color_idx != NULL) {
+ *color_idx = inverse_color_order[color_map[r * cols + c]];
+ }
return color_ctx;
}
#endif // CONFIG_PALETTE
diff --git a/av1/common/entropymode.h b/av1/common/entropymode.h
index 85c68e1..68a6400 100644
--- a/av1/common/entropymode.h
+++ b/av1/common/entropymode.h
@@ -359,7 +359,8 @@
#if CONFIG_PALETTE
int av1_get_palette_color_context(const uint8_t *color_map, int cols, int r,
- int c, int n, int *color_order);
+ int c, int n, uint8_t *color_order,
+ int *color_idx);
#endif // CONFIG_PALETTE
#ifdef __cplusplus
diff --git a/av1/decoder/detokenize.c b/av1/decoder/detokenize.c
index 42679da..f2f74f5 100644
--- a/av1/decoder/detokenize.c
+++ b/av1/decoder/detokenize.c
@@ -297,15 +297,15 @@
#if CONFIG_PALETTE
void av1_decode_palette_tokens(MACROBLOCKD *const xd, int plane,
aom_reader *r) {
- MODE_INFO *const mi = xd->mi[0];
- MB_MODE_INFO *const mbmi = &mi->mbmi;
+ const MODE_INFO *const mi = xd->mi[0];
+ const MB_MODE_INFO *const mbmi = &mi->mbmi;
const BLOCK_SIZE bsize = mbmi->sb_type;
const int rows = (4 * num_4x4_blocks_high_lookup[bsize]) >>
(xd->plane[plane != 0].subsampling_y);
const int cols = (4 * num_4x4_blocks_wide_lookup[bsize]) >>
(xd->plane[plane != 0].subsampling_x);
- int color_idx, color_ctx, color_order[PALETTE_MAX_SIZE];
- int n = mbmi->palette_mode_info.palette_size[plane != 0];
+ uint8_t color_order[PALETTE_MAX_SIZE];
+ const int n = mbmi->palette_mode_info.palette_size[plane != 0];
int i, j;
uint8_t *color_map = xd->plane[plane != 0].color_index_map;
const aom_prob(*const prob)[PALETTE_COLOR_CONTEXTS][PALETTE_COLORS - 1] =
@@ -314,10 +314,10 @@
for (i = 0; i < rows; ++i) {
for (j = (i == 0 ? 1 : 0); j < cols; ++j) {
- color_ctx =
- av1_get_palette_color_context(color_map, cols, i, j, n, color_order);
- color_idx = aom_read_tree(r, av1_palette_color_tree[n - 2],
- prob[n - 2][color_ctx], ACCT_STR);
+ const int color_ctx = av1_get_palette_color_context(color_map, cols, i, j,
+ n, color_order, NULL);
+ const int color_idx = aom_read_tree(r, av1_palette_color_tree[n - 2],
+ prob[n - 2][color_ctx], ACCT_STR);
assert(color_idx >= 0 && color_idx < n);
color_map[i * cols + j] = color_order[color_idx];
}
diff --git a/av1/encoder/dct.c b/av1/encoder/dct.c
index 63b71a5..221e3cd 100644
--- a/av1/encoder/dct.c
+++ b/av1/encoder/dct.c
@@ -325,7 +325,6 @@
range_check(output, 16, 16);
}
-#if CONFIG_EXT_TX
static void fdct32(const tran_low_t *input, tran_low_t *output) {
tran_high_t temp;
tran_low_t step[32];
@@ -723,7 +722,6 @@
range_check(output, 32, 18);
}
-#endif // CONFIG_EXT_TX
static void fadst4(const tran_low_t *input, tran_low_t *output) {
tran_high_t x0, x1, x2, x3;
@@ -1809,57 +1807,74 @@
}
#endif // CONFIG_AOM_HIGHBITDEPTH
-#if CONFIG_EXT_TX
+// TODO(luoyi): Adding this function to avoid DCT_DCT overflow.
+// Remove this function after we scale the column txfm output correctly.
+static INLINE int range_check_dct32x32(const int16_t *input, int16_t bound,
+ int size) {
+ int i;
+ for (i = 0; i < size; ++i) {
+ if (abs(input[i]) > bound) return 1;
+ }
+ return 0;
+}
+
void av1_fht32x32_c(const int16_t *input, tran_low_t *output, int stride,
int tx_type) {
- if (tx_type == DCT_DCT) {
- aom_fdct32x32_c(input, output, stride);
- } else {
- static const transform_2d FHT[] = {
- { fdct32, fdct32 }, // DCT_DCT
- { fhalfright32, fdct32 }, // ADST_DCT
- { fdct32, fhalfright32 }, // DCT_ADST
- { fhalfright32, fhalfright32 }, // ADST_ADST
- { fhalfright32, fdct32 }, // FLIPADST_DCT
- { fdct32, fhalfright32 }, // DCT_FLIPADST
- { fhalfright32, fhalfright32 }, // FLIPADST_FLIPADST
- { fhalfright32, fhalfright32 }, // ADST_FLIPADST
- { fhalfright32, fhalfright32 }, // FLIPADST_ADST
- { fidtx32, fidtx32 }, // IDTX
- { fdct32, fidtx32 }, // V_DCT
- { fidtx32, fdct32 }, // H_DCT
- { fhalfright32, fidtx32 }, // V_ADST
- { fidtx32, fhalfright32 }, // H_ADST
- { fhalfright32, fidtx32 }, // V_FLIPADST
- { fidtx32, fhalfright32 }, // H_FLIPADST
- };
- const transform_2d ht = FHT[tx_type];
- tran_low_t out[1024];
- int i, j;
- tran_low_t temp_in[32], temp_out[32];
+ static const transform_2d FHT[] = {
+ { fdct32, fdct32 }, // DCT_DCT
+#if CONFIG_EXT_TX
+ { fhalfright32, fdct32 }, // ADST_DCT
+ { fdct32, fhalfright32 }, // DCT_ADST
+ { fhalfright32, fhalfright32 }, // ADST_ADST
+ { fhalfright32, fdct32 }, // FLIPADST_DCT
+ { fdct32, fhalfright32 }, // DCT_FLIPADST
+ { fhalfright32, fhalfright32 }, // FLIPADST_FLIPADST
+ { fhalfright32, fhalfright32 }, // ADST_FLIPADST
+ { fhalfright32, fhalfright32 }, // FLIPADST_ADST
+ { fidtx32, fidtx32 }, // IDTX
+ { fdct32, fidtx32 }, // V_DCT
+ { fidtx32, fdct32 }, // H_DCT
+ { fhalfright32, fidtx32 }, // V_ADST
+ { fidtx32, fhalfright32 }, // H_ADST
+ { fhalfright32, fidtx32 }, // V_FLIPADST
+ { fidtx32, fhalfright32 }, // H_FLIPADST
+#endif
+ };
+ const transform_2d ht = FHT[tx_type];
+ tran_low_t out[1024];
+ int i, j;
+ tran_low_t temp_in[32], temp_out[32];
- int16_t flipped_input[32 * 32];
- maybe_flip_input(&input, &stride, 32, 32, flipped_input, tx_type);
+#if CONFIG_EXT_TX
+ int16_t flipped_input[32 * 32];
+ maybe_flip_input(&input, &stride, 32, 32, flipped_input, tx_type);
+#endif
- // Columns
- for (i = 0; i < 32; ++i) {
- for (j = 0; j < 32; ++j) temp_in[j] = input[j * stride + i] * 4;
- ht.cols(temp_in, temp_out);
- for (j = 0; j < 32; ++j)
- out[j * 32 + i] = (temp_out[j] + 1 + (temp_out[j] > 0)) >> 2;
+ if (DCT_DCT == tx_type) {
+ if (range_check_dct32x32(input, (1 << 6) - 1, 1 << 10)) {
+ aom_fdct32x32_c(input, output, stride);
+ return;
}
+ }
+ // Columns
+ for (i = 0; i < 32; ++i) {
+ for (j = 0; j < 32; ++j) temp_in[j] = input[j * stride + i] * 4;
+ ht.cols(temp_in, temp_out);
+ for (j = 0; j < 32; ++j)
+ out[j * 32 + i] = (temp_out[j] + 1 + (temp_out[j] > 0)) >> 2;
+ }
- // Rows
- for (i = 0; i < 32; ++i) {
- for (j = 0; j < 32; ++j) temp_in[j] = out[j + i * 32];
- ht.rows(temp_in, temp_out);
- for (j = 0; j < 32; ++j)
- output[j + i * 32] =
- (tran_low_t)((temp_out[j] + 1 + (temp_out[j] < 0)) >> 2);
- }
+ // Rows
+ for (i = 0; i < 32; ++i) {
+ for (j = 0; j < 32; ++j) temp_in[j] = out[j + i * 32];
+ ht.rows(temp_in, temp_out);
+ for (j = 0; j < 32; ++j)
+ output[j + i * 32] =
+ (tran_low_t)((temp_out[j] + 1 + (temp_out[j] < 0)) >> 2);
}
}
+#if CONFIG_EXT_TX
// Forward identity transform.
void av1_fwd_idtx_c(const int16_t *src_diff, tran_low_t *coeff, int stride,
int bs, int tx_type) {
diff --git a/av1/encoder/encoder.c b/av1/encoder/encoder.c
index c22c5a8..f1a6f72 100644
--- a/av1/encoder/encoder.c
+++ b/av1/encoder/encoder.c
@@ -422,39 +422,6 @@
aom_free(cpi->segmentation_map);
cpi->segmentation_map = NULL;
-#if CONFIG_REF_MV
- for (i = 0; i < NMV_CONTEXTS; ++i) {
- aom_free(cpi->nmv_costs[i][0]);
- aom_free(cpi->nmv_costs[i][1]);
- aom_free(cpi->nmv_costs_hp[i][0]);
- aom_free(cpi->nmv_costs_hp[i][1]);
- cpi->nmv_costs[i][0] = NULL;
- cpi->nmv_costs[i][1] = NULL;
- cpi->nmv_costs_hp[i][0] = NULL;
- cpi->nmv_costs_hp[i][1] = NULL;
- }
-#endif
-
- aom_free(cpi->nmvcosts[0]);
- aom_free(cpi->nmvcosts[1]);
- cpi->nmvcosts[0] = NULL;
- cpi->nmvcosts[1] = NULL;
-
- aom_free(cpi->nmvcosts_hp[0]);
- aom_free(cpi->nmvcosts_hp[1]);
- cpi->nmvcosts_hp[0] = NULL;
- cpi->nmvcosts_hp[1] = NULL;
-
- aom_free(cpi->nmvsadcosts[0]);
- aom_free(cpi->nmvsadcosts[1]);
- cpi->nmvsadcosts[0] = NULL;
- cpi->nmvsadcosts[1] = NULL;
-
- aom_free(cpi->nmvsadcosts_hp[0]);
- aom_free(cpi->nmvsadcosts_hp[1]);
- cpi->nmvsadcosts_hp[0] = NULL;
- cpi->nmvsadcosts_hp[1] = NULL;
-
av1_cyclic_refresh_free(cpi->cyclic_refresh);
cpi->cyclic_refresh = NULL;
@@ -512,27 +479,15 @@
#if CONFIG_REF_MV
for (i = 0; i < NMV_CONTEXTS; ++i) {
av1_copy(cc->nmv_vec_cost[i], cpi->td.mb.nmv_vec_cost[i]);
- memcpy(cc->nmv_costs[i][0], cpi->nmv_costs[i][0],
- MV_VALS * sizeof(*cpi->nmv_costs[i][0]));
- memcpy(cc->nmv_costs[i][1], cpi->nmv_costs[i][1],
- MV_VALS * sizeof(*cpi->nmv_costs[i][1]));
- memcpy(cc->nmv_costs_hp[i][0], cpi->nmv_costs_hp[i][0],
- MV_VALS * sizeof(*cpi->nmv_costs_hp[i][0]));
- memcpy(cc->nmv_costs_hp[i][1], cpi->nmv_costs_hp[i][1],
- MV_VALS * sizeof(*cpi->nmv_costs_hp[i][1]));
+ av1_copy(cc->nmv_costs, cpi->nmv_costs);
+ av1_copy(cc->nmv_costs_hp, cpi->nmv_costs_hp);
}
#else
av1_copy(cc->nmvjointcost, cpi->td.mb.nmvjointcost);
#endif
- memcpy(cc->nmvcosts[0], cpi->nmvcosts[0],
- MV_VALS * sizeof(*cpi->nmvcosts[0]));
- memcpy(cc->nmvcosts[1], cpi->nmvcosts[1],
- MV_VALS * sizeof(*cpi->nmvcosts[1]));
- memcpy(cc->nmvcosts_hp[0], cpi->nmvcosts_hp[0],
- MV_VALS * sizeof(*cpi->nmvcosts_hp[0]));
- memcpy(cc->nmvcosts_hp[1], cpi->nmvcosts_hp[1],
- MV_VALS * sizeof(*cpi->nmvcosts_hp[1]));
+ av1_copy(cc->nmvcosts, cpi->nmvcosts);
+ av1_copy(cc->nmvcosts_hp, cpi->nmvcosts_hp);
av1_copy(cc->last_ref_lf_deltas, cm->lf.last_ref_deltas);
av1_copy(cc->last_mode_lf_deltas, cm->lf.last_mode_deltas);
@@ -552,25 +507,15 @@
#if CONFIG_REF_MV
for (i = 0; i < NMV_CONTEXTS; ++i) {
av1_copy(cpi->td.mb.nmv_vec_cost[i], cc->nmv_vec_cost[i]);
- memcpy(cpi->nmv_costs[i][0], cc->nmv_costs[i][0],
- MV_VALS * sizeof(*cc->nmv_costs[i][0]));
- memcpy(cpi->nmv_costs[i][1], cc->nmv_costs[i][1],
- MV_VALS * sizeof(*cc->nmv_costs[i][1]));
- memcpy(cpi->nmv_costs_hp[i][0], cc->nmv_costs_hp[i][0],
- MV_VALS * sizeof(*cc->nmv_costs_hp[i][0]));
- memcpy(cpi->nmv_costs_hp[i][1], cc->nmv_costs_hp[i][1],
- MV_VALS * sizeof(*cc->nmv_costs_hp[i][1]));
+ av1_copy(cpi->nmv_costs, cc->nmv_costs);
+ av1_copy(cpi->nmv_costs_hp, cc->nmv_costs_hp);
}
#else
av1_copy(cpi->td.mb.nmvjointcost, cc->nmvjointcost);
#endif
- memcpy(cpi->nmvcosts[0], cc->nmvcosts[0], MV_VALS * sizeof(*cc->nmvcosts[0]));
- memcpy(cpi->nmvcosts[1], cc->nmvcosts[1], MV_VALS * sizeof(*cc->nmvcosts[1]));
- memcpy(cpi->nmvcosts_hp[0], cc->nmvcosts_hp[0],
- MV_VALS * sizeof(*cc->nmvcosts_hp[0]));
- memcpy(cpi->nmvcosts_hp[1], cc->nmvcosts_hp[1],
- MV_VALS * sizeof(*cc->nmvcosts_hp[1]));
+ av1_copy(cpi->nmvcosts, cc->nmvcosts);
+ av1_copy(cpi->nmvcosts_hp, cc->nmvcosts_hp);
av1_copy(cm->lf.last_ref_deltas, cc->last_ref_lf_deltas);
av1_copy(cm->lf.last_mode_deltas, cc->last_mode_lf_deltas);
@@ -2117,33 +2062,15 @@
#if CONFIG_REF_MV
for (i = 0; i < NMV_CONTEXTS; ++i) {
- CHECK_MEM_ERROR(cm, cpi->nmv_costs[i][0],
- aom_calloc(MV_VALS, sizeof(*cpi->nmv_costs[i][0])));
- CHECK_MEM_ERROR(cm, cpi->nmv_costs[i][1],
- aom_calloc(MV_VALS, sizeof(*cpi->nmv_costs[i][1])));
- CHECK_MEM_ERROR(cm, cpi->nmv_costs_hp[i][0],
- aom_calloc(MV_VALS, sizeof(*cpi->nmv_costs_hp[i][0])));
- CHECK_MEM_ERROR(cm, cpi->nmv_costs_hp[i][1],
- aom_calloc(MV_VALS, sizeof(*cpi->nmv_costs_hp[i][1])));
+ memset(cpi->nmv_costs, 0, sizeof(cpi->nmv_costs));
+ memset(cpi->nmv_costs_hp, 0, sizeof(cpi->nmv_costs_hp));
}
#endif
- CHECK_MEM_ERROR(cm, cpi->nmvcosts[0],
- aom_calloc(MV_VALS, sizeof(*cpi->nmvcosts[0])));
- CHECK_MEM_ERROR(cm, cpi->nmvcosts[1],
- aom_calloc(MV_VALS, sizeof(*cpi->nmvcosts[1])));
- CHECK_MEM_ERROR(cm, cpi->nmvcosts_hp[0],
- aom_calloc(MV_VALS, sizeof(*cpi->nmvcosts_hp[0])));
- CHECK_MEM_ERROR(cm, cpi->nmvcosts_hp[1],
- aom_calloc(MV_VALS, sizeof(*cpi->nmvcosts_hp[1])));
- CHECK_MEM_ERROR(cm, cpi->nmvsadcosts[0],
- aom_calloc(MV_VALS, sizeof(*cpi->nmvsadcosts[0])));
- CHECK_MEM_ERROR(cm, cpi->nmvsadcosts[1],
- aom_calloc(MV_VALS, sizeof(*cpi->nmvsadcosts[1])));
- CHECK_MEM_ERROR(cm, cpi->nmvsadcosts_hp[0],
- aom_calloc(MV_VALS, sizeof(*cpi->nmvsadcosts_hp[0])));
- CHECK_MEM_ERROR(cm, cpi->nmvsadcosts_hp[1],
- aom_calloc(MV_VALS, sizeof(*cpi->nmvsadcosts_hp[1])));
+ memset(cpi->nmvcosts, 0, sizeof(cpi->nmvcosts));
+ memset(cpi->nmvcosts_hp, 0, sizeof(cpi->nmvcosts_hp));
+ memset(cpi->nmvsadcosts, 0, sizeof(cpi->nmvsadcosts));
+ memset(cpi->nmvsadcosts_hp, 0, sizeof(cpi->nmvsadcosts_hp));
for (i = 0; i < (sizeof(cpi->mbgraph_stats) / sizeof(cpi->mbgraph_stats[0]));
i++) {
diff --git a/av1/encoder/encoder.h b/av1/encoder/encoder.h
index b55481b..0c66905 100644
--- a/av1/encoder/encoder.h
+++ b/av1/encoder/encoder.h
@@ -414,14 +414,14 @@
CODING_CONTEXT coding_context;
#if CONFIG_REF_MV
- int *nmv_costs[NMV_CONTEXTS][2];
- int *nmv_costs_hp[NMV_CONTEXTS][2];
+ int nmv_costs[NMV_CONTEXTS][2][MV_VALS];
+ int nmv_costs_hp[NMV_CONTEXTS][2][MV_VALS];
#endif
- int *nmvcosts[2];
- int *nmvcosts_hp[2];
- int *nmvsadcosts[2];
- int *nmvsadcosts_hp[2];
+ int nmvcosts[2][MV_VALS];
+ int nmvcosts_hp[2][MV_VALS];
+ int nmvsadcosts[2][MV_VALS];
+ int nmvsadcosts_hp[2][MV_VALS];
int64_t last_time_stamp_seen;
int64_t last_end_time_stamp_seen;
diff --git a/av1/encoder/hybrid_fwd_txfm.c b/av1/encoder/hybrid_fwd_txfm.c
index 1103c4b..6d5eccd 100644
--- a/av1/encoder/hybrid_fwd_txfm.c
+++ b/av1/encoder/hybrid_fwd_txfm.c
@@ -21,7 +21,7 @@
if (rd_transform)
aom_fdct32x32_rd(src, dst, src_stride);
else
- aom_fdct32x32(src, dst, src_stride);
+ av1_fht32x32(src, dst, src_stride, DCT_DCT);
}
static void fwd_txfm_4x4(const int16_t *src_diff, tran_low_t *coeff,
diff --git a/av1/encoder/rdopt.c b/av1/encoder/rdopt.c
index 8d151a7..8ba6b7b 100644
--- a/av1/encoder/rdopt.c
+++ b/av1/encoder/rdopt.c
@@ -1771,8 +1771,7 @@
if (colors > 1 && colors <= 64) {
int r, c, i, j, k;
const int max_itr = 50;
- int color_ctx, color_idx = 0;
- int color_order[PALETTE_MAX_SIZE];
+ uint8_t color_order[PALETTE_MAX_SIZE];
float *const data = x->palette_buffer->kmeans_data_buf;
float centroids[PALETTE_MAX_SIZE];
uint8_t *const color_map = xd->plane[0].color_index_map;
@@ -1856,13 +1855,9 @@
1);
for (i = 0; i < rows; ++i) {
for (j = (i == 0 ? 1 : 0); j < cols; ++j) {
- color_ctx = av1_get_palette_color_context(color_map, cols, i, j, k,
- color_order);
- for (r = 0; r < k; ++r)
- if (color_map[i * cols + j] == color_order[r]) {
- color_idx = r;
- break;
- }
+ int color_idx;
+ const int color_ctx = av1_get_palette_color_context(
+ color_map, cols, i, j, k, color_order, &color_idx);
assert(color_idx >= 0 && color_idx < k);
this_rate += cpi->palette_y_color_cost[k - 2][color_ctx][color_idx];
}
@@ -2507,7 +2502,7 @@
static void angle_estimation(const uint8_t *src, int src_stride, int rows,
int cols, uint8_t *directional_mode_skip_mask) {
- int i, r, c, dx, dy, temp, sn, remd, quot;
+ int i, r, c, index, dx, dy, temp, sn, remd, quot;
uint64_t hist[DIRECTIONAL_MODES];
uint64_t hist_sum = 0;
@@ -2515,7 +2510,6 @@
src += src_stride;
for (r = 1; r < rows; ++r) {
for (c = 1; c < cols; ++c) {
- uint8_t index;
dx = src[c] - src[c - 1];
dy = src[c] - src[c - src_stride];
temp = dx * dx + dy * dy;
@@ -2538,16 +2532,16 @@
for (i = 0; i < DIRECTIONAL_MODES; ++i) hist_sum += hist[i];
for (i = 0; i < INTRA_MODES; ++i) {
if (i != DC_PRED && i != TM_PRED) {
- const uint8_t index = mode_to_angle_bin[i];
- uint64_t score = 2 * hist[index];
+ const uint8_t angle_bin = mode_to_angle_bin[i];
+ uint64_t score = 2 * hist[angle_bin];
int weight = 2;
- if (index > 0) {
- score += hist[index - 1];
- weight += 1;
+ if (angle_bin > 0) {
+ score += hist[angle_bin - 1];
+ ++weight;
}
- if (index < DIRECTIONAL_MODES - 1) {
- score += hist[index + 1];
- weight += 1;
+ if (angle_bin < DIRECTIONAL_MODES - 1) {
+ score += hist[angle_bin + 1];
+ ++weight;
}
if (score * ANGLE_SKIP_THRESH < hist_sum * weight)
directional_mode_skip_mask[i] = 1;
@@ -2559,7 +2553,7 @@
static void highbd_angle_estimation(const uint8_t *src8, int src_stride,
int rows, int cols,
uint8_t *directional_mode_skip_mask) {
- int i, r, c, dx, dy, temp, sn, remd, quot;
+ int i, r, c, index, dx, dy, temp, sn, remd, quot;
uint64_t hist[DIRECTIONAL_MODES];
uint64_t hist_sum = 0;
uint16_t *src = CONVERT_TO_SHORTPTR(src8);
@@ -2568,7 +2562,6 @@
src += src_stride;
for (r = 1; r < rows; ++r) {
for (c = 1; c < cols; ++c) {
- uint8_t index;
dx = src[c] - src[c - 1];
dy = src[c] - src[c - src_stride];
temp = dx * dx + dy * dy;
@@ -2591,16 +2584,16 @@
for (i = 0; i < DIRECTIONAL_MODES; ++i) hist_sum += hist[i];
for (i = 0; i < INTRA_MODES; ++i) {
if (i != DC_PRED && i != TM_PRED) {
- const uint8_t index = mode_to_angle_bin[i];
- uint64_t score = 2 * hist[index];
+ const uint8_t angle_bin = mode_to_angle_bin[i];
+ uint64_t score = 2 * hist[angle_bin];
int weight = 2;
- if (index > 0) {
- score += hist[index - 1];
- weight += 1;
+ if (angle_bin > 0) {
+ score += hist[angle_bin - 1];
+ ++weight;
}
- if (index < DIRECTIONAL_MODES - 1) {
- score += hist[index + 1];
- weight += 1;
+ if (angle_bin < DIRECTIONAL_MODES - 1) {
+ score += hist[angle_bin + 1];
+ ++weight;
}
if (score * ANGLE_SKIP_THRESH < hist_sum * weight)
directional_mode_skip_mask[i] = 1;
@@ -3654,8 +3647,7 @@
if (colors > 1 && colors <= 64) {
int r, c, n, i, j;
const int max_itr = 50;
- int color_ctx, color_idx = 0;
- int color_order[PALETTE_MAX_SIZE];
+ uint8_t color_order[PALETTE_MAX_SIZE];
int64_t this_sse;
float lb_u, ub_u, val_u;
float lb_v, ub_v, val_v;
@@ -3748,13 +3740,9 @@
for (i = 0; i < rows; ++i) {
for (j = (i == 0 ? 1 : 0); j < cols; ++j) {
- color_ctx = av1_get_palette_color_context(color_map, cols, i, j, n,
- color_order);
- for (r = 0; r < n; ++r)
- if (color_map[i * cols + j] == color_order[r]) {
- color_idx = r;
- break;
- }
+ int color_idx;
+ const int color_ctx = av1_get_palette_color_context(
+ color_map, cols, i, j, n, color_order, &color_idx);
assert(color_idx >= 0 && color_idx < n);
this_rate += cpi->palette_uv_color_cost[n - 2][color_ctx][color_idx];
}
@@ -9385,7 +9373,7 @@
int best_rate_nocoef;
#endif
int64_t distortion2 = 0, distortion_y = 0, dummy_rd = best_rd, this_rd;
- int skippable = 0;
+ int skippable = 0, rate_overhead = 0;
TX_SIZE best_tx_size, uv_tx;
TX_TYPE best_tx_type;
PALETTE_MODE_INFO palette_mode_info;
diff --git a/av1/encoder/tokenize.c b/av1/encoder/tokenize.c
index 13abe6b..67f4b5d 100644
--- a/av1/encoder/tokenize.c
+++ b/av1/encoder/tokenize.c
@@ -410,18 +410,19 @@
}
#if CONFIG_PALETTE
-void av1_tokenize_palette_sb(const AV1_COMP *cpi, struct ThreadData *const td,
- int plane, TOKENEXTRA **t, RUN_TYPE dry_run,
- BLOCK_SIZE bsize, int *rate) {
- MACROBLOCK *const x = &td->mb;
- MACROBLOCKD *const xd = &x->e_mbd;
- MB_MODE_INFO *mbmi = &xd->mi[0]->mbmi;
- uint8_t *color_map = xd->plane[plane != 0].color_index_map;
- PALETTE_MODE_INFO *pmi = &mbmi->palette_mode_info;
- int n = pmi->palette_size[plane != 0];
- int i, j, k;
+void av1_tokenize_palette_sb(const AV1_COMP *cpi,
+ const struct ThreadData *const td, int plane,
+ TOKENEXTRA **t, RUN_TYPE dry_run, BLOCK_SIZE bsize,
+ int *rate) {
+ const MACROBLOCK *const x = &td->mb;
+ const MACROBLOCKD *const xd = &x->e_mbd;
+ const MB_MODE_INFO *const mbmi = &xd->mi[0]->mbmi;
+ const uint8_t *const color_map = xd->plane[plane != 0].color_index_map;
+ const PALETTE_MODE_INFO *const pmi = &mbmi->palette_mode_info;
+ const int n = pmi->palette_size[plane != 0];
+ int i, j;
int this_rate = 0;
- int color_idx = -1, color_ctx, color_order[PALETTE_MAX_SIZE];
+ uint8_t color_order[PALETTE_MAX_SIZE];
const int rows = (4 * num_4x4_blocks_high_lookup[bsize]) >>
(xd->plane[plane != 0].subsampling_y);
const int cols = (4 * num_4x4_blocks_wide_lookup[bsize]) >>
@@ -432,17 +433,13 @@
for (i = 0; i < rows; ++i) {
for (j = (i == 0 ? 1 : 0); j < cols; ++j) {
- color_ctx =
- av1_get_palette_color_context(color_map, cols, i, j, n, color_order);
- for (k = 0; k < n; ++k)
- if (color_map[i * cols + j] == color_order[k]) {
- color_idx = k;
- break;
- }
- assert(color_idx >= 0 && color_idx < n);
+ int color_new_idx;
+ const int color_ctx = av1_get_palette_color_context(
+ color_map, cols, i, j, n, color_order, &color_new_idx);
+ assert(color_new_idx >= 0 && color_new_idx < n);
if (dry_run == DRY_RUN_COSTCOEFFS)
- this_rate += cpi->palette_y_color_cost[n - 2][color_ctx][color_idx];
- (*t)->token = color_idx;
+ this_rate += cpi->palette_y_color_cost[n - 2][color_ctx][color_new_idx];
+ (*t)->token = color_new_idx;
(*t)->context_tree = probs[n - 2][color_ctx];
(*t)->skip_eob_node = 0;
++(*t);
diff --git a/av1/encoder/tokenize.h b/av1/encoder/tokenize.h
index ae896a6..89610df 100644
--- a/av1/encoder/tokenize.h
+++ b/av1/encoder/tokenize.h
@@ -72,7 +72,7 @@
#endif
#if CONFIG_PALETTE
void av1_tokenize_palette_sb(const struct AV1_COMP *cpi,
- struct ThreadData *const td, int plane,
+ const struct ThreadData *const td, int plane,
TOKENEXTRA **t, RUN_TYPE dry_run, BLOCK_SIZE bsize,
int *rate);
#endif // CONFIG_PALETTE
diff --git a/av1/encoder/x86/hybrid_fwd_txfm_avx2.c b/av1/encoder/x86/hybrid_fwd_txfm_avx2.c
index 69bf89a..928af13 100644
--- a/av1/encoder/x86/hybrid_fwd_txfm_avx2.c
+++ b/av1/encoder/x86/hybrid_fwd_txfm_avx2.c
@@ -198,8 +198,8 @@
in[15] = _mm256_permute2x128_si256(tr0_7, tr0_f, 0x31);
}
-static void load_buffer_16x16(const int16_t *input, int stride, int flipud,
- int fliplr, __m256i *in) {
+static INLINE void load_buffer_16x16(const int16_t *input, int stride,
+ int flipud, int fliplr, __m256i *in) {
if (!flipud) {
in[0] = _mm256_loadu_si256((const __m256i *)(input + 0 * stride));
in[1] = _mm256_loadu_si256((const __m256i *)(input + 1 * stride));
@@ -1273,7 +1273,6 @@
_mm256_zeroupper();
}
-#if CONFIG_EXT_TX
static void mm256_vectors_swap(__m256i *a0, __m256i *a1, const int size) {
int i = 0;
__m256i temp;
@@ -1622,7 +1621,6 @@
mm256_transpose_32x32(in0, in1);
}
-#endif // CONFIG_EXT_TX
static INLINE void write_buffer_32x32(const __m256i *in0, const __m256i *in1,
int stride, tran_low_t *output) {
@@ -1667,9 +1665,11 @@
mm256_vectors_swap(in1, &in1[16], 16);
mm256_transpose_32x32(in0, in1);
}
+#endif // CONFIG_EXT_TX
-static void load_buffer_32x32(const int16_t *input, int stride, int flipud,
- int fliplr, __m256i *in0, __m256i *in1) {
+static INLINE void load_buffer_32x32(const int16_t *input, int stride,
+ int flipud, int fliplr, __m256i *in0,
+ __m256i *in1) {
// Load 4 16x16 blocks
const int16_t *topL = input;
const int16_t *topR = input + 16;
@@ -1708,7 +1708,6 @@
load_buffer_16x16(topR, stride, flipud, fliplr, in1);
load_buffer_16x16(botR, stride, flipud, fliplr, in1 + 16);
}
-#endif // CONFIG_EXT_TX
static void nr_right_shift_32x32_16col(__m256i *in) {
int i = 0;
@@ -1729,8 +1728,7 @@
nr_right_shift_32x32_16col(in1);
}
-#if CONFIG_EXT_TX
-static void pr_right_shift_32x32_16col(__m256i *in) {
+static INLINE void pr_right_shift_32x32_16col(__m256i *in) {
int i = 0;
const __m256i zero = _mm256_setzero_si256();
const __m256i one = _mm256_set1_epi16(1);
@@ -1745,11 +1743,12 @@
}
// Positive rounding
-static void pr_right_shift_32x32(__m256i *in0, __m256i *in1) {
+static INLINE void pr_right_shift_32x32(__m256i *in0, __m256i *in1) {
pr_right_shift_32x32_16col(in0);
pr_right_shift_32x32_16col(in1);
}
+#if CONFIG_EXT_TX
static void fidtx32_avx2(__m256i *in0, __m256i *in1) {
int i = 0;
while (i < 32) {
@@ -1761,23 +1760,42 @@
}
#endif
+static INLINE int range_check_dct32x32(const __m256i *in0, const __m256i *in1,
+ int row) {
+ __m256i value, bits0, bits1;
+ const __m256i bound = _mm256_set1_epi16((1 << 6) - 1);
+ int flag;
+ int i = 0;
+
+ while (i < row) {
+ value = _mm256_abs_epi16(in0[i]);
+ bits0 = _mm256_cmpgt_epi16(value, bound);
+ value = _mm256_abs_epi16(in1[i]);
+ bits1 = _mm256_cmpgt_epi16(value, bound);
+ bits0 = _mm256_or_si256(bits0, bits1);
+ flag = _mm256_movemask_epi8(bits0);
+ if (flag) return 1;
+ i++;
+ }
+ return 0;
+}
+
void av1_fht32x32_avx2(const int16_t *input, tran_low_t *output, int stride,
int tx_type) {
__m256i in0[32]; // left 32 columns
__m256i in1[32]; // right 32 columns
- (void)input;
- (void)stride;
switch (tx_type) {
-// TODO(luoyi): For DCT_DCT, fwd_txfm_32x32() uses aom set. But this
-// function has better speed. The replacement must work with the
-// corresponding inverse transform.
-// case DCT_DCT:
-// load_buffer_32x32(input, stride, 0, 0, in0, in1);
-// fdct32_avx2(in0, in1);
-// pr_right_shift_32x32(in0, in1);
-// fdct32_avx2(in0, in1);
-// break;
+ case DCT_DCT:
+ load_buffer_32x32(input, stride, 0, 0, in0, in1);
+ if (range_check_dct32x32(in0, in1, 32)) {
+ aom_fdct32x32_avx2(input, output, stride);
+ return;
+ }
+ fdct32_avx2(in0, in1);
+ pr_right_shift_32x32(in0, in1);
+ fdct32_avx2(in0, in1);
+ break;
#if CONFIG_EXT_TX
case ADST_DCT:
load_buffer_32x32(input, stride, 0, 0, in0, in1);
diff --git a/configure b/configure
index 99d2bb8..2659d37 100755
--- a/configure
+++ b/configure
@@ -616,20 +616,18 @@
check_add_cflags -Wvla
check_add_cflags -Wimplicit-function-declaration
check_add_cflags -Wuninitialized
- check_add_cflags -Wunused-variable
+ check_add_cflags -Wunused
check_add_cflags -Wsign-compare
+ # Enabling the following warning (in combination with -Wunused above)
+ # for C++ generates errors in third_party code including googletest and
+ # libyuv. So enable it only for C code.
+ check_cflags "-Wextra" && add_cflags_only "-Wextra"
# Enabling the following warning for C++ generates some useless warnings
# about some function parameters shadowing class member function names.
# So, only enable this warning for C code.
check_cflags "-Wshadow" && add_cflags_only "-Wshadow"
- case ${CC} in
- *clang*) ;;
- *) check_add_cflags -Wunused-but-set-variable ;;
- esac
if enabled mips || [ -z "${INLINE}" ]; then
enabled extra_warnings || check_add_cflags -Wno-unused-function
- else
- check_add_cflags -Wunused-function
fi
fi
diff --git a/examples/aom_cx_set_ref.c b/examples/aom_cx_set_ref.c
index fdb9739..6beb4fb 100644
--- a/examples/aom_cx_set_ref.c
+++ b/examples/aom_cx_set_ref.c
@@ -191,8 +191,7 @@
}
static void testing_decode(aom_codec_ctx_t *encoder, aom_codec_ctx_t *decoder,
- aom_codec_enc_cfg_t *cfg, unsigned int frame_out,
- int *mismatch_seen) {
+ unsigned int frame_out, int *mismatch_seen) {
aom_image_t enc_img, dec_img;
struct av1_ref_frame ref_enc, ref_dec;
@@ -226,11 +225,10 @@
aom_img_free(&dec_img);
}
-static int encode_frame(aom_codec_ctx_t *ecodec, aom_codec_enc_cfg_t *cfg,
- aom_image_t *img, unsigned int frame_in,
- AvxVideoWriter *writer, int test_decode,
- aom_codec_ctx_t *dcodec, unsigned int *frame_out,
- int *mismatch_seen) {
+static int encode_frame(aom_codec_ctx_t *ecodec, aom_image_t *img,
+ unsigned int frame_in, AvxVideoWriter *writer,
+ int test_decode, aom_codec_ctx_t *dcodec,
+ unsigned int *frame_out, int *mismatch_seen) {
int got_pkts = 0;
aom_codec_iter_t iter = NULL;
const aom_codec_cx_pkt_t *pkt = NULL;
@@ -271,7 +269,7 @@
// Mismatch checking
if (got_data && test_decode) {
- testing_decode(ecodec, dcodec, cfg, *frame_out, mismatch_seen);
+ testing_decode(ecodec, dcodec, *frame_out, mismatch_seen);
}
return got_pkts;
@@ -280,12 +278,12 @@
int main(int argc, char **argv) {
FILE *infile = NULL;
// Encoder
- aom_codec_ctx_t ecodec = { 0 };
- aom_codec_enc_cfg_t cfg = { 0 };
+ aom_codec_ctx_t ecodec;
+ aom_codec_enc_cfg_t cfg;
unsigned int frame_in = 0;
aom_image_t raw;
aom_codec_err_t res;
- AvxVideoInfo info = { 0 };
+ AvxVideoInfo info;
AvxVideoWriter *writer = NULL;
const AvxInterface *encoder = NULL;
@@ -311,6 +309,12 @@
unsigned int limit = 0;
exec_name = argv[0];
+ // Clear explicitly, as simply assigning "{ 0 }" generates
+ // "missing-field-initializers" warning in some compilers.
+ memset(&ecodec, 0, sizeof(ecodec));
+ memset(&cfg, 0, sizeof(cfg));
+ memset(&info, 0, sizeof(info));
+
if (argc < 7) die("Invalid number of arguments");
codec_arg = argv[1];
@@ -404,7 +408,7 @@
}
}
- encode_frame(&ecodec, &cfg, &raw, frame_in, writer, test_decode, &dcodec,
+ encode_frame(&ecodec, &raw, frame_in, writer, test_decode, &dcodec,
&frame_out, &mismatch_seen);
frame_in++;
if (mismatch_seen) break;
@@ -412,8 +416,8 @@
// Flush encoder.
if (!mismatch_seen)
- while (encode_frame(&ecodec, &cfg, NULL, frame_in, writer, test_decode,
- &dcodec, &frame_out, &mismatch_seen)) {
+ while (encode_frame(&ecodec, NULL, frame_in, writer, test_decode, &dcodec,
+ &frame_out, &mismatch_seen)) {
}
printf("\n");
diff --git a/examples/lossless_encoder.c b/examples/lossless_encoder.c
index 069e35e..1abeb27 100644
--- a/examples/lossless_encoder.c
+++ b/examples/lossless_encoder.c
@@ -63,13 +63,17 @@
int frame_count = 0;
aom_image_t raw;
aom_codec_err_t res;
- AvxVideoInfo info = { 0 };
+ AvxVideoInfo info;
AvxVideoWriter *writer = NULL;
const AvxInterface *encoder = NULL;
const int fps = 30;
exec_name = argv[0];
+ // Clear explicitly, as simply assigning "{ 0 }" generates
+ // "missing-field-initializers" warning in some compilers.
+ memset(&info, 0, sizeof(info));
+
if (argc < 5) die("Invalid number of arguments");
encoder = get_aom_encoder_by_name("av1");
diff --git a/examples/simple_encoder.c b/examples/simple_encoder.c
index 418757d..1d2b51e 100644
--- a/examples/simple_encoder.c
+++ b/examples/simple_encoder.c
@@ -151,7 +151,7 @@
int frame_count = 0;
aom_image_t raw;
aom_codec_err_t res;
- AvxVideoInfo info = { 0 };
+ AvxVideoInfo info;
AvxVideoWriter *writer = NULL;
const AvxInterface *encoder = NULL;
const int fps = 30;
@@ -168,6 +168,10 @@
exec_name = argv[0];
+ // Clear explicitly, as simply assigning "{ 0 }" generates
+ // "missing-field-initializers" warning in some compilers.
+ memset(&info, 0, sizeof(info));
+
if (argc != 9) die("Invalid number of arguments");
codec_arg = argv[1];
diff --git a/test/av1_convolve_optimz_test.cc b/test/av1_convolve_optimz_test.cc
index b83ae94..b891e99 100644
--- a/test/av1_convolve_optimz_test.cc
+++ b/test/av1_convolve_optimz_test.cc
@@ -54,7 +54,6 @@
const size_t maxBlockSize = maxWidth * maxHeight;
const int horizOffset = 32;
const int vertiOffset = 32;
-const size_t testMaxBlk = 128;
const int stride = 128;
const int x_step_q4 = 16;
@@ -90,7 +89,7 @@
void RunVertFilterBitExactCheck();
private:
- void PrepFilterBuffer(int w, int h);
+ void PrepFilterBuffer();
void DiffFilterBuffer();
conv_filter_t conv_horiz_;
conv_filter_t conv_vert_;
@@ -106,7 +105,7 @@
int avg_;
};
-void AV1ConvolveOptimzTest::PrepFilterBuffer(int w, int h) {
+void AV1ConvolveOptimzTest::PrepFilterBuffer() {
int r, c;
ACMRandom rnd(ACMRandom::DeterministicSeed());
@@ -150,7 +149,7 @@
}
void AV1ConvolveOptimzTest::RunHorizFilterBitExactCheck() {
- PrepFilterBuffer(testMaxBlk, testMaxBlk);
+ PrepFilterBuffer();
InterpFilterParams filter_params = av1_get_interp_filter_params(filter_);
@@ -167,7 +166,7 @@
// and test again.
int intermediate_height =
(((height_ - 1) * 16 + subpel_) >> SUBPEL_BITS) + filter_params.taps;
- PrepFilterBuffer(testMaxBlk, testMaxBlk);
+ PrepFilterBuffer();
av1_convolve_horiz_c(src_ref_, stride, dst_ref_, stride, width_,
intermediate_height, filter_params, subpel_, x_step_q4,
@@ -180,7 +179,7 @@
}
void AV1ConvolveOptimzTest::RunVertFilterBitExactCheck() {
- PrepFilterBuffer(testMaxBlk, testMaxBlk);
+ PrepFilterBuffer();
InterpFilterParams filter_params = av1_get_interp_filter_params(filter_);
@@ -266,7 +265,7 @@
void RunVertFilterBitExactCheck();
private:
- void PrepFilterBuffer(int w, int h);
+ void PrepFilterBuffer();
void DiffFilterBuffer();
hbd_conv_filter_t conv_horiz_;
hbd_conv_filter_t conv_vert_;
@@ -283,7 +282,7 @@
int bit_depth_;
};
-void AV1HbdConvolveOptimzTest::PrepFilterBuffer(int w, int h) {
+void AV1HbdConvolveOptimzTest::PrepFilterBuffer() {
int r, c;
ACMRandom rnd(ACMRandom::DeterministicSeed());
@@ -326,7 +325,7 @@
}
void AV1HbdConvolveOptimzTest::RunHorizFilterBitExactCheck() {
- PrepFilterBuffer(testMaxBlk, testMaxBlk);
+ PrepFilterBuffer();
InterpFilterParams filter_params = av1_get_interp_filter_params(filter_);
@@ -344,7 +343,7 @@
// and test again.
int intermediate_height =
(((height_ - 1) * 16 + subpel_) >> SUBPEL_BITS) + filter_params.taps;
- PrepFilterBuffer(testMaxBlk, testMaxBlk);
+ PrepFilterBuffer();
av1_highbd_convolve_horiz_c(src_, stride, dst_ref_, stride, width_,
intermediate_height, filter_params, subpel_,
@@ -357,7 +356,7 @@
}
void AV1HbdConvolveOptimzTest::RunVertFilterBitExactCheck() {
- PrepFilterBuffer(testMaxBlk, testMaxBlk);
+ PrepFilterBuffer();
InterpFilterParams filter_params = av1_get_interp_filter_params(filter_);
diff --git a/test/av1_dct_test.cc b/test/av1_dct_test.cc
index ac1a551..d5c23f6 100644
--- a/test/av1_dct_test.cc
+++ b/test/av1_dct_test.cc
@@ -102,5 +102,6 @@
C, AV1FwdTxfm,
::testing::Values(FdctParam(&fdct4, &reference_dct_1d, 4, 1),
FdctParam(&fdct8, &reference_dct_1d, 8, 1),
- FdctParam(&fdct16, &reference_dct_1d, 16, 2)));
+ FdctParam(&fdct16, &reference_dct_1d, 16, 2),
+ FdctParam(&fdct32, &reference_dct_1d, 32, 3)));
} // namespace
diff --git a/test/codec_factory.h b/test/codec_factory.h
index c92d5c1..b645102 100644
--- a/test/codec_factory.h
+++ b/test/codec_factory.h
@@ -123,6 +123,9 @@
#if CONFIG_AV1_DECODER
return new AV1Decoder(cfg, flags, deadline);
#else
+ (void)cfg;
+ (void)flags;
+ (void)deadline;
return NULL;
#endif
}
@@ -134,6 +137,10 @@
#if CONFIG_AV1_ENCODER
return new AV1Encoder(cfg, deadline, init_flags, stats);
#else
+ (void)cfg;
+ (void)deadline;
+ (void)init_flags;
+ (void)stats;
return NULL;
#endif
}
@@ -143,6 +150,8 @@
#if CONFIG_AV1_ENCODER
return aom_codec_enc_config_default(&aom_codec_av1_cx_algo, cfg, usage);
#else
+ (void)cfg;
+ (void)usage;
return AOM_CODEC_INCAPABLE;
#endif
}
diff --git a/test/dct16x16_test.cc b/test/dct16x16_test.cc
index e73daa5..9811955 100644
--- a/test/dct16x16_test.cc
+++ b/test/dct16x16_test.cc
@@ -264,12 +264,12 @@
}
void idct16x16_10_ref(const tran_low_t *in, uint8_t *out, int stride,
- int tx_type) {
+ int /*tx_type*/) {
idct16x16_10(in, out, stride);
}
void idct16x16_12_ref(const tran_low_t *in, uint8_t *out, int stride,
- int tx_type) {
+ int /*tx_type*/) {
idct16x16_12(in, out, stride);
}
@@ -727,7 +727,7 @@
virtual void TearDown() { libaom_test::ClearSystemState(); }
protected:
- void RunFwdTxfm(int16_t *in, tran_low_t *out, int stride) {}
+ void RunFwdTxfm(int16_t * /*in*/, tran_low_t * /*out*/, int /*stride*/) {}
void RunInvTxfm(tran_low_t *out, uint8_t *dst, int stride) {
inv_txfm_(out, dst, stride);
}
diff --git a/test/decode_test_driver.cc b/test/decode_test_driver.cc
index 95a0eb5..7adb9d6 100644
--- a/test/decode_test_driver.cc
+++ b/test/decode_test_driver.cc
@@ -92,7 +92,7 @@
aom_codec_err_t res_dec =
decoder->DecodeFrame(video->cxdata(), video->frame_size());
- if (!HandleDecodeResult(res_dec, *video, decoder)) break;
+ if (!HandleDecodeResult(res_dec, decoder)) break;
} else {
// Signal end of the file to the decoder.
const aom_codec_err_t res_dec = decoder->DecodeFrame(NULL, 0);
diff --git a/test/decode_test_driver.h b/test/decode_test_driver.h
index aabca40..b8f8d1a 100644
--- a/test/decode_test_driver.h
+++ b/test/decode_test_driver.h
@@ -141,7 +141,6 @@
// Hook to be called to handle decode result. Return true to continue.
virtual bool HandleDecodeResult(const aom_codec_err_t res_dec,
- const CompressedVideoSource & /*video*/,
Decoder *decoder) {
EXPECT_EQ(AOM_CODEC_OK, res_dec) << decoder->DecodeError();
return AOM_CODEC_OK == res_dec;
diff --git a/test/encode_test_driver.cc b/test/encode_test_driver.cc
index c1a0cb7..092e669 100644
--- a/test/encode_test_driver.cc
+++ b/test/encode_test_driver.cc
@@ -275,7 +275,7 @@
aom_codec_err_t res_dec = decoder->DecodeFrame(
(const uint8_t *)pkt->data.frame.buf, pkt->data.frame.sz);
- if (!HandleDecodeResult(res_dec, *video, decoder.get())) break;
+ if (!HandleDecodeResult(res_dec, decoder.get())) break;
has_dxdata = true;
}
@@ -293,7 +293,7 @@
// Flush the decoder when there are no more fragments.
if ((init_flags_ & AOM_CODEC_USE_OUTPUT_PARTITION) && has_dxdata) {
const aom_codec_err_t res_dec = decoder->DecodeFrame(NULL, 0);
- if (!HandleDecodeResult(res_dec, *video, decoder.get())) break;
+ if (!HandleDecodeResult(res_dec, decoder.get())) break;
}
if (has_dxdata && has_cxdata) {
diff --git a/test/encode_test_driver.h b/test/encode_test_driver.h
index 11c387a..45a080e 100644
--- a/test/encode_test_driver.h
+++ b/test/encode_test_driver.h
@@ -228,7 +228,6 @@
// Hook to be called to handle decode result. Return true to continue.
virtual bool HandleDecodeResult(const aom_codec_err_t res_dec,
- const VideoSource & /*video*/,
Decoder *decoder) {
EXPECT_EQ(AOM_CODEC_OK, res_dec) << decoder->DecodeError();
return AOM_CODEC_OK == res_dec;
diff --git a/test/encoder_parms_get_to_decoder.cc b/test/encoder_parms_get_to_decoder.cc
index 640e12f..52d68b1 100644
--- a/test/encoder_parms_get_to_decoder.cc
+++ b/test/encoder_parms_get_to_decoder.cc
@@ -94,7 +94,6 @@
}
virtual bool HandleDecodeResult(const aom_codec_err_t res_dec,
- const libaom_test::VideoSource & /*video*/,
libaom_test::Decoder *decoder) {
aom_codec_ctx_t *const av1_decoder = decoder->GetDecoder();
aom_codec_alg_priv_t *const priv =
diff --git a/test/error_resilience_test.cc b/test/error_resilience_test.cc
index 951c47f..07b6039 100644
--- a/test/error_resilience_test.cc
+++ b/test/error_resilience_test.cc
@@ -55,8 +55,7 @@
nframes_++;
}
- virtual void PreEncodeFrameHook(libaom_test::VideoSource *video,
- ::libaom_test::Encoder * /*encoder*/) {
+ virtual void PreEncodeFrameHook(libaom_test::VideoSource *video) {
frame_flags_ &=
~(AOM_EFLAG_NO_UPD_LAST | AOM_EFLAG_NO_UPD_GF | AOM_EFLAG_NO_UPD_ARF);
if (droppable_nframes_ > 0 &&
diff --git a/test/fht32x32_test.cc b/test/fht32x32_test.cc
index a949ebf..3d07b44 100644
--- a/test/fht32x32_test.cc
+++ b/test/fht32x32_test.cc
@@ -69,6 +69,7 @@
inv_txfm_ = GET_PARAM(1);
tx_type_ = GET_PARAM(2);
pitch_ = 32;
+ height_ = 32;
fwd_txfm_ref = fht32x32_ref;
bit_depth_ = GET_PARAM(3);
mask_ = (1 << bit_depth_) - 1;
@@ -90,6 +91,7 @@
};
TEST_P(AV1Trans32x32HT, CoeffCheck) { RunCoeffCheck(); }
+TEST_P(AV1Trans32x32HT, MemCheck) { RunMemCheck(); }
#if CONFIG_AOM_HIGHBITDEPTH
class AV1HighbdTrans32x32HT
@@ -164,8 +166,7 @@
#if HAVE_AVX2
const Ht32x32Param kArrayHt32x32Param_avx2[] = {
- // TODO(luoyi): DCT_DCT tx_type is not enabled in av1_fht32x32_c(avx2) yet.
- // make_tuple(&av1_fht32x32_avx2, dummy_inv_txfm, 0, AOM_BITS_8, 1024),
+ make_tuple(&av1_fht32x32_avx2, dummy_inv_txfm, 0, AOM_BITS_8, 1024),
make_tuple(&av1_fht32x32_avx2, dummy_inv_txfm, 1, AOM_BITS_8, 1024),
make_tuple(&av1_fht32x32_avx2, dummy_inv_txfm, 2, AOM_BITS_8, 1024),
make_tuple(&av1_fht32x32_avx2, dummy_inv_txfm, 3, AOM_BITS_8, 1024),
diff --git a/test/frame_size_tests.cc b/test/frame_size_tests.cc
index f1fad70..25b8718 100644
--- a/test/frame_size_tests.cc
+++ b/test/frame_size_tests.cc
@@ -28,7 +28,6 @@
}
virtual bool HandleDecodeResult(const aom_codec_err_t res_dec,
- const libaom_test::VideoSource & /*video*/,
libaom_test::Decoder *decoder) {
EXPECT_EQ(expected_res_, res_dec) << decoder->DecodeError();
return !::testing::Test::HasFailure();
diff --git a/test/variance_test.cc b/test/variance_test.cc
index 5ff5090..7848e20 100644
--- a/test/variance_test.cc
+++ b/test/variance_test.cc
@@ -101,8 +101,7 @@
}
RoundHighBitDepth(bit_depth, &se, &sse);
*sse_ptr = static_cast<uint32_t>(sse);
- return static_cast<uint32_t>(
- sse - ((static_cast<int64_t>(se) * se) >> (l2w + l2h)));
+ return static_cast<uint32_t>(sse - ((se * se) >> (l2w + l2h)));
}
/* The subpel reference functions differ from the codec version in one aspect:
@@ -157,8 +156,7 @@
}
RoundHighBitDepth(bit_depth, &se, &sse);
*sse_ptr = static_cast<uint32_t>(sse);
- return static_cast<uint32_t>(
- sse - ((static_cast<int64_t>(se) * se) >> (l2w + l2h)));
+ return static_cast<uint32_t>(sse - ((se * se) >> (l2w + l2h)));
}
static uint32_t subpel_avg_variance_ref(const uint8_t *ref, const uint8_t *src,
@@ -211,8 +209,7 @@
}
RoundHighBitDepth(bit_depth, &se, &sse);
*sse_ptr = static_cast<uint32_t>(sse);
- return static_cast<uint32_t>(
- sse - ((static_cast<int64_t>(se) * se) >> (l2w + l2h)));
+ return static_cast<uint32_t>(sse - ((se * se) >> (l2w + l2h)));
}
////////////////////////////////////////////////////////////////////////////////
diff --git a/webmenc.cc b/webmenc.cc
index f78f027..e3d209a 100644
--- a/webmenc.cc
+++ b/webmenc.cc
@@ -24,7 +24,6 @@
void write_webm_file_header(struct WebmOutputContext *webm_ctx,
const aom_codec_enc_cfg_t *cfg,
- const struct aom_rational *fps,
stereo_format_t stereo_fmt, unsigned int fourcc,
const struct AvxRational *par) {
mkvmuxer::MkvWriter *const writer = new mkvmuxer::MkvWriter(webm_ctx->stream);
diff --git a/webmenc.h b/webmenc.h
index 90211ff..74387fb 100644
--- a/webmenc.h
+++ b/webmenc.h
@@ -40,7 +40,6 @@
void write_webm_file_header(struct WebmOutputContext *webm_ctx,
const aom_codec_enc_cfg_t *cfg,
- const struct aom_rational *fps,
stereo_format_t stereo_fmt, unsigned int fourcc,
const struct AvxRational *par);
