kenmirrors/freeswitch
1
0
Fork 0
mirror of https://github.com/signalwire/freeswitch.git synced 2025-08-13 01:26:58 +00:00
Code Issues Projects Releases Wiki Activity

update to libvpx b46243d from repo https://chromium.googlesource.com/webm/libvpx

Browse Source
This commit is contained in:
Mike Jerris
2016-09-26 14:31:51 -04:00
parent 2b1f0da5c4
commit 8c5f0301f3
1008 changed files with 110177 additions and 201326 deletions
Download Patch File Download Diff File Expand all files Collapse all files

270
libs/libvpx/vpx_dsp/vpx_convolve.c
View File

@@ -21,8 +21,8 @@
static void convolve_horiz(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const InterpKernel *x_filters,
int x0_q4, int x_step_q4, int w, int h) {
const InterpKernel *x_filters, int x0_q4,
int x_step_q4, int w, int h) {
int x, y;
src -= SUBPEL_TAPS / 2 - 1;
for (y = 0; y < h; ++y) {
@@ -31,8 +31,7 @@ static void convolve_horiz(const uint8_t *src, ptrdiff_t src_stride,
const uint8_t *const src_x = &src[x_q4 >> SUBPEL_BITS];
const int16_t *const x_filter = x_filters[x_q4 & SUBPEL_MASK];
int k, sum = 0;
for (k = 0; k < SUBPEL_TAPS; ++k)
sum += src_x[k] * x_filter[k];
for (k = 0; k < SUBPEL_TAPS; ++k) sum += src_x[k] * x_filter[k];
dst[x] = clip_pixel(ROUND_POWER_OF_TWO(sum, FILTER_BITS));
x_q4 += x_step_q4;
}
@@ -43,8 +42,8 @@ static void convolve_horiz(const uint8_t *src, ptrdiff_t src_stride,
static void convolve_avg_horiz(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const InterpKernel *x_filters,
int x0_q4, int x_step_q4, int w, int h) {
const InterpKernel *x_filters, int x0_q4,
int x_step_q4, int w, int h) {
int x, y;
src -= SUBPEL_TAPS / 2 - 1;
for (y = 0; y < h; ++y) {
@@ -53,10 +52,9 @@ static void convolve_avg_horiz(const uint8_t *src, ptrdiff_t src_stride,
const uint8_t *const src_x = &src[x_q4 >> SUBPEL_BITS];
const int16_t *const x_filter = x_filters[x_q4 & SUBPEL_MASK];
int k, sum = 0;
for (k = 0; k < SUBPEL_TAPS; ++k)
sum += src_x[k] * x_filter[k];
dst[x] = ROUND_POWER_OF_TWO(dst[x] +
clip_pixel(ROUND_POWER_OF_TWO(sum, FILTER_BITS)), 1);
for (k = 0; k < SUBPEL_TAPS; ++k) sum += src_x[k] * x_filter[k];
dst[x] = ROUND_POWER_OF_TWO(
dst[x] + clip_pixel(ROUND_POWER_OF_TWO(sum, FILTER_BITS)), 1);
x_q4 += x_step_q4;
}
src += src_stride;
@@ -66,8 +64,8 @@ static void convolve_avg_horiz(const uint8_t *src, ptrdiff_t src_stride,
static void convolve_vert(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const InterpKernel *y_filters,
int y0_q4, int y_step_q4, int w, int h) {
const InterpKernel *y_filters, int y0_q4,
int y_step_q4, int w, int h) {
int x, y;
src -= src_stride * (SUBPEL_TAPS / 2 - 1);
@@ -89,8 +87,8 @@ static void convolve_vert(const uint8_t *src, ptrdiff_t src_stride,
static void convolve_avg_vert(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const InterpKernel *y_filters,
int y0_q4, int y_step_q4, int w, int h) {
const InterpKernel *y_filters, int y0_q4,
int y_step_q4, int w, int h) {
int x, y;
src -= src_stride * (SUBPEL_TAPS / 2 - 1);
@@ -102,8 +100,10 @@ static void convolve_avg_vert(const uint8_t *src, ptrdiff_t src_stride,
int k, sum = 0;
for (k = 0; k < SUBPEL_TAPS; ++k)
sum += src_y[k * src_stride] * y_filter[k];
dst[y * dst_stride] = ROUND_POWER_OF_TWO(dst[y * dst_stride] +
clip_pixel(ROUND_POWER_OF_TWO(sum, FILTER_BITS)), 1);
dst[y * dst_stride] = ROUND_POWER_OF_TWO(
dst[y * dst_stride] +
clip_pixel(ROUND_POWER_OF_TWO(sum, FILTER_BITS)),
1);
y_q4 += y_step_q4;
}
++src;
@@ -111,13 +111,11 @@ static void convolve_avg_vert(const uint8_t *src, ptrdiff_t src_stride,
}
}
static void convolve(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const InterpKernel *const x_filters,
static void convolve(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst,
ptrdiff_t dst_stride, const InterpKernel *const x_filters,
int x0_q4, int x_step_q4,
const InterpKernel *const y_filters,
int y0_q4, int y_step_q4,
int w, int h) {
const InterpKernel *const y_filters, int y0_q4,
int y_step_q4, int w, int h) {
// Note: Fixed size intermediate buffer, temp, places limits on parameters.
// 2d filtering proceeds in 2 steps:
// (1) Interpolate horizontally into an intermediate buffer, temp.
@@ -132,7 +130,7 @@ static void convolve(const uint8_t *src, ptrdiff_t src_stride,
// --((64 - 1) * 32 + 15) >> 4 + 8 = 135.
uint8_t temp[135 * 64];
int intermediate_height =
(((h - 1) * y_step_q4 + y0_q4) >> SUBPEL_BITS) + SUBPEL_TAPS;
(((h - 1) * y_step_q4 + y0_q4) >> SUBPEL_BITS) + SUBPEL_TAPS;
assert(w <= 64);
assert(h <= 64);
@@ -158,67 +156,66 @@ static int get_filter_offset(const int16_t *f, const InterpKernel *base) {
void vpx_convolve8_horiz_c(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) {
const int16_t *filter_y, int y_step_q4, int w,
int h) {
const InterpKernel *const filters_x = get_filter_base(filter_x);
const int x0_q4 = get_filter_offset(filter_x, filters_x);
(void)filter_y;
(void)y_step_q4;
convolve_horiz(src, src_stride, dst, dst_stride, filters_x,
x0_q4, x_step_q4, w, h);
convolve_horiz(src, src_stride, dst, dst_stride, filters_x, x0_q4, x_step_q4,
w, h);
}
void vpx_convolve8_avg_horiz_c(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) {
const int16_t *filter_y, int y_step_q4, int w,
int h) {
const InterpKernel *const filters_x = get_filter_base(filter_x);
const int x0_q4 = get_filter_offset(filter_x, filters_x);
(void)filter_y;
(void)y_step_q4;
convolve_avg_horiz(src, src_stride, dst, dst_stride, filters_x,
x0_q4, x_step_q4, w, h);
convolve_avg_horiz(src, src_stride, dst, dst_stride, filters_x, x0_q4,
x_step_q4, w, h);
}
void vpx_convolve8_vert_c(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) {
const int16_t *filter_y, int y_step_q4, int w,
int h) {
const InterpKernel *const filters_y = get_filter_base(filter_y);
const int y0_q4 = get_filter_offset(filter_y, filters_y);
(void)filter_x;
(void)x_step_q4;
convolve_vert(src, src_stride, dst, dst_stride, filters_y,
y0_q4, y_step_q4, w, h);
convolve_vert(src, src_stride, dst, dst_stride, filters_y, y0_q4, y_step_q4,
w, h);
}
void vpx_convolve8_avg_vert_c(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) {
const int16_t *filter_y, int y_step_q4, int w,
int h) {
const InterpKernel *const filters_y = get_filter_base(filter_y);
const int y0_q4 = get_filter_offset(filter_y, filters_y);
(void)filter_x;
(void)x_step_q4;
convolve_avg_vert(src, src_stride, dst, dst_stride, filters_y,
y0_q4, y_step_q4, w, h);
convolve_avg_vert(src, src_stride, dst, dst_stride, filters_y, y0_q4,
y_step_q4, w, h);
}
void vpx_convolve8_c(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,
void vpx_convolve8_c(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) {
const InterpKernel *const filters_x = get_filter_base(filter_x);
const int x0_q4 = get_filter_offset(filter_x, filters_x);
@@ -226,35 +223,34 @@ void vpx_convolve8_c(const uint8_t *src, ptrdiff_t src_stride,
const InterpKernel *const filters_y = get_filter_base(filter_y);
const int y0_q4 = get_filter_offset(filter_y, filters_y);
convolve(src, src_stride, dst, dst_stride,
filters_x, x0_q4, x_step_q4,
convolve(src, src_stride, dst, dst_stride, filters_x, x0_q4, x_step_q4,
filters_y, y0_q4, y_step_q4, w, h);
}
void vpx_convolve8_avg_c(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,
void vpx_convolve8_avg_c(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) {
/* Fixed size intermediate buffer places limits on parameters. */
DECLARE_ALIGNED(16, uint8_t, temp[64 * 64]);
assert(w <= 64);
assert(h <= 64);
vpx_convolve8_c(src, src_stride, temp, 64,
filter_x, x_step_q4, filter_y, y_step_q4, w, h);
vpx_convolve8_c(src, src_stride, temp, 64, filter_x, x_step_q4, filter_y,
y_step_q4, w, h);
vpx_convolve_avg_c(temp, 64, dst, dst_stride, NULL, 0, NULL, 0, w, h);
}
void vpx_convolve_copy_c(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const int16_t *filter_x, int filter_x_stride,
const int16_t *filter_y, int filter_y_stride,
int w, int h) {
void vpx_convolve_copy_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst,
ptrdiff_t dst_stride, const int16_t *filter_x,
int filter_x_stride, const int16_t *filter_y,
int filter_y_stride, int w, int h) {
int r;
(void)filter_x; (void)filter_x_stride;
(void)filter_y; (void)filter_y_stride;
(void)filter_x;
(void)filter_x_stride;
(void)filter_y;
(void)filter_y_stride;
for (r = h; r > 0; --r) {
memcpy(dst, src, w);
@@ -263,47 +259,44 @@ void vpx_convolve_copy_c(const uint8_t *src, ptrdiff_t src_stride,
}
}
void vpx_convolve_avg_c(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const int16_t *filter_x, int filter_x_stride,
const int16_t *filter_y, int filter_y_stride,
int w, int h) {
void vpx_convolve_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst,
ptrdiff_t dst_stride, const int16_t *filter_x,
int filter_x_stride, const int16_t *filter_y,
int filter_y_stride, int w, int h) {
int x, y;
(void)filter_x; (void)filter_x_stride;
(void)filter_y; (void)filter_y_stride;
(void)filter_x;
(void)filter_x_stride;
(void)filter_y;
(void)filter_y_stride;
for (y = 0; y < h; ++y) {
for (x = 0; x < w; ++x)
dst[x] = ROUND_POWER_OF_TWO(dst[x] + src[x], 1);
for (x = 0; x < w; ++x) dst[x] = ROUND_POWER_OF_TWO(dst[x] + src[x], 1);
src += src_stride;
dst += dst_stride;
}
}
void vpx_scaled_horiz_c(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,
void vpx_scaled_horiz_c(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) {
vpx_convolve8_horiz_c(src, src_stride, dst, dst_stride, filter_x, x_step_q4,
filter_y, y_step_q4, w, h);
}
void vpx_scaled_vert_c(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,
void vpx_scaled_vert_c(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) {
vpx_convolve8_vert_c(src, src_stride, dst, dst_stride, filter_x, x_step_q4,
filter_y, y_step_q4, w, h);
}
void vpx_scaled_2d_c(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,
void vpx_scaled_2d_c(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) {
vpx_convolve8_c(src, src_stride, dst, dst_stride, filter_x, x_step_q4,
filter_y, y_step_q4, w, h);
@@ -312,8 +305,8 @@ void vpx_scaled_2d_c(const uint8_t *src, ptrdiff_t src_stride,
void vpx_scaled_avg_horiz_c(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) {
const int16_t *filter_y, int y_step_q4, int w,
int h) {
vpx_convolve8_avg_horiz_c(src, src_stride, dst, dst_stride, filter_x,
x_step_q4, filter_y, y_step_q4, w, h);
}
@@ -321,17 +314,16 @@ void vpx_scaled_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride,
void vpx_scaled_avg_vert_c(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) {
const int16_t *filter_y, int y_step_q4, int w,
int h) {
vpx_convolve8_avg_vert_c(src, src_stride, dst, dst_stride, filter_x,
x_step_q4, filter_y, y_step_q4, w, h);
}
void vpx_scaled_avg_2d_c(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 vpx_scaled_avg_2d_c(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) {
vpx_convolve8_avg_c(src, src_stride, dst, dst_stride, filter_x, x_step_q4,
filter_y, y_step_q4, w, h);
}
@@ -339,9 +331,8 @@ void vpx_scaled_avg_2d_c(const uint8_t *src, ptrdiff_t src_stride,
#if CONFIG_VP9_HIGHBITDEPTH
static void highbd_convolve_horiz(const uint8_t *src8, ptrdiff_t src_stride,
uint8_t *dst8, ptrdiff_t dst_stride,
const InterpKernel *x_filters,
int x0_q4, int x_step_q4,
int w, int h, int bd) {
const InterpKernel *x_filters, int x0_q4,
int x_step_q4, int w, int h, int bd) {
int x, y;
uint16_t *src = CONVERT_TO_SHORTPTR(src8);
uint16_t *dst = CONVERT_TO_SHORTPTR(dst8);
@@ -352,8 +343,7 @@ static void highbd_convolve_horiz(const uint8_t *src8, ptrdiff_t src_stride,
const uint16_t *const src_x = &src[x_q4 >> SUBPEL_BITS];
const int16_t *const x_filter = x_filters[x_q4 & SUBPEL_MASK];
int k, sum = 0;
for (k = 0; k < SUBPEL_TAPS; ++k)
sum += src_x[k] * x_filter[k];
for (k = 0; k < SUBPEL_TAPS; ++k) sum += src_x[k] * x_filter[k];
dst[x] = clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
x_q4 += x_step_q4;
}
@@ -364,9 +354,8 @@ static void highbd_convolve_horiz(const uint8_t *src8, ptrdiff_t src_stride,
static void highbd_convolve_avg_horiz(const uint8_t *src8, ptrdiff_t src_stride,
uint8_t *dst8, ptrdiff_t dst_stride,
const InterpKernel *x_filters,
int x0_q4, int x_step_q4,
int w, int h, int bd) {
const InterpKernel *x_filters, int x0_q4,
int x_step_q4, int w, int h, int bd) {
int x, y;
uint16_t *src = CONVERT_TO_SHORTPTR(src8);
uint16_t *dst = CONVERT_TO_SHORTPTR(dst8);
@@ -377,10 +366,10 @@ static void highbd_convolve_avg_horiz(const uint8_t *src8, ptrdiff_t src_stride,
const uint16_t *const src_x = &src[x_q4 >> SUBPEL_BITS];
const int16_t *const x_filter = x_filters[x_q4 & SUBPEL_MASK];
int k, sum = 0;
for (k = 0; k < SUBPEL_TAPS; ++k)
sum += src_x[k] * x_filter[k];
dst[x] = ROUND_POWER_OF_TWO(dst[x] +
clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd), 1);
for (k = 0; k < SUBPEL_TAPS; ++k) sum += src_x[k] * x_filter[k];
dst[x] = ROUND_POWER_OF_TWO(
dst[x] + clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd),
1);
x_q4 += x_step_q4;
}
src += src_stride;
@@ -390,9 +379,8 @@ static void highbd_convolve_avg_horiz(const uint8_t *src8, ptrdiff_t src_stride,
static void highbd_convolve_vert(const uint8_t *src8, ptrdiff_t src_stride,
uint8_t *dst8, ptrdiff_t dst_stride,
const InterpKernel *y_filters,
int y0_q4, int y_step_q4, int w, int h,
int bd) {
const InterpKernel *y_filters, int y0_q4,
int y_step_q4, int w, int h, int bd) {
int x, y;
uint16_t *src = CONVERT_TO_SHORTPTR(src8);
uint16_t *dst = CONVERT_TO_SHORTPTR(dst8);
@@ -405,8 +393,8 @@ static void highbd_convolve_vert(const uint8_t *src8, ptrdiff_t src_stride,
int k, sum = 0;
for (k = 0; k < SUBPEL_TAPS; ++k)
sum += src_y[k * src_stride] * y_filter[k];
dst[y * dst_stride] = clip_pixel_highbd(
ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
dst[y * dst_stride] =
clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
y_q4 += y_step_q4;
}
++src;
@@ -416,9 +404,8 @@ static void highbd_convolve_vert(const uint8_t *src8, ptrdiff_t src_stride,
static void highbd_convolve_avg_vert(const uint8_t *src8, ptrdiff_t src_stride,
uint8_t *dst8, ptrdiff_t dst_stride,
const InterpKernel *y_filters,
int y0_q4, int y_step_q4, int w, int h,
int bd) {
const InterpKernel *y_filters, int y0_q4,
int y_step_q4, int w, int h, int bd) {
int x, y;
uint16_t *src = CONVERT_TO_SHORTPTR(src8);
uint16_t *dst = CONVERT_TO_SHORTPTR(dst8);
@@ -431,8 +418,10 @@ static void highbd_convolve_avg_vert(const uint8_t *src8, ptrdiff_t src_stride,
int k, sum = 0;
for (k = 0; k < SUBPEL_TAPS; ++k)
sum += src_y[k * src_stride] * y_filter[k];
dst[y * dst_stride] = ROUND_POWER_OF_TWO(dst[y * dst_stride] +
clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd), 1);
dst[y * dst_stride] = ROUND_POWER_OF_TWO(
dst[y * dst_stride] +
clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd),
1);
y_q4 += y_step_q4;
}
++src;
@@ -442,11 +431,9 @@ static void highbd_convolve_avg_vert(const uint8_t *src8, ptrdiff_t src_stride,
static void highbd_convolve(const uint8_t *src, ptrdiff_t src_stride,
uint8_t *dst, ptrdiff_t dst_stride,
const InterpKernel *const x_filters,
int x0_q4, int x_step_q4,
const InterpKernel *const y_filters,
int y0_q4, int y_step_q4,
int w, int h, int bd) {
const InterpKernel *const x_filters, int x0_q4,
int x_step_q4, const InterpKernel *const y_filters,
int y0_q4, int y_step_q4, int w, int h, int bd) {
// Note: Fixed size intermediate buffer, temp, places limits on parameters.
// 2d filtering proceeds in 2 steps:
// (1) Interpolate horizontally into an intermediate buffer, temp.
@@ -461,35 +448,33 @@ static void highbd_convolve(const uint8_t *src, ptrdiff_t src_stride,
// --((64 - 1) * 32 + 15) >> 4 + 8 = 135.
uint16_t temp[64 * 135];
int intermediate_height =
(((h - 1) * y_step_q4 + y0_q4) >> SUBPEL_BITS) + SUBPEL_TAPS;
(((h - 1) * y_step_q4 + y0_q4) >> SUBPEL_BITS) + SUBPEL_TAPS;
assert(w <= 64);
assert(h <= 64);
assert(y_step_q4 <= 32);
assert(x_step_q4 <= 32);
highbd_convolve_horiz(src - src_stride * (SUBPEL_TAPS / 2 - 1),
src_stride, CONVERT_TO_BYTEPTR(temp), 64,
x_filters, x0_q4, x_step_q4, w,
intermediate_height, bd);
highbd_convolve_horiz(src - src_stride * (SUBPEL_TAPS / 2 - 1), src_stride,
CONVERT_TO_BYTEPTR(temp), 64, x_filters, x0_q4,
x_step_q4, w, intermediate_height, bd);
highbd_convolve_vert(CONVERT_TO_BYTEPTR(temp) + 64 * (SUBPEL_TAPS / 2 - 1),
64, dst, dst_stride, y_filters, y0_q4, y_step_q4,
w, h, bd);
64, dst, dst_stride, y_filters, y0_q4, y_step_q4, w, h,
bd);
}
void vpx_highbd_convolve8_horiz_c(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, int bd) {
const int16_t *filter_y, int y_step_q4, int w,
int h, int bd) {
const InterpKernel *const filters_x = get_filter_base(filter_x);
const int x0_q4 = get_filter_offset(filter_x, filters_x);
(void)filter_y;
(void)y_step_q4;
highbd_convolve_horiz(src, src_stride, dst, dst_stride, filters_x,
x0_q4, x_step_q4, w, h, bd);
highbd_convolve_horiz(src, src_stride, dst, dst_stride, filters_x, x0_q4,
x_step_q4, w, h, bd);
}
void vpx_highbd_convolve8_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride,
@@ -502,22 +487,22 @@ void vpx_highbd_convolve8_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride,
(void)filter_y;
(void)y_step_q4;
highbd_convolve_avg_horiz(src, src_stride, dst, dst_stride, filters_x,
x0_q4, x_step_q4, w, h, bd);
highbd_convolve_avg_horiz(src, src_stride, dst, dst_stride, filters_x, x0_q4,
x_step_q4, w, h, bd);
}
void vpx_highbd_convolve8_vert_c(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, int bd) {
const int16_t *filter_y, int y_step_q4, int w,
int h, int bd) {
const InterpKernel *const filters_y = get_filter_base(filter_y);
const int y0_q4 = get_filter_offset(filter_y, filters_y);
(void)filter_x;
(void)x_step_q4;
highbd_convolve_vert(src, src_stride, dst, dst_stride, filters_y,
y0_q4, y_step_q4, w, h, bd);
highbd_convolve_vert(src, src_stride, dst, dst_stride, filters_y, y0_q4,
y_step_q4, w, h, bd);
}
void vpx_highbd_convolve8_avg_vert_c(const uint8_t *src, ptrdiff_t src_stride,
@@ -530,31 +515,30 @@ void vpx_highbd_convolve8_avg_vert_c(const uint8_t *src, ptrdiff_t src_stride,
(void)filter_x;
(void)x_step_q4;
highbd_convolve_avg_vert(src, src_stride, dst, dst_stride, filters_y,
y0_q4, y_step_q4, w, h, bd);
highbd_convolve_avg_vert(src, src_stride, dst, dst_stride, filters_y, y0_q4,
y_step_q4, w, h, bd);
}
void vpx_highbd_convolve8_c(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, int bd) {
const int16_t *filter_y, int y_step_q4, int w,
int h, int bd) {
const InterpKernel *const filters_x = get_filter_base(filter_x);
const int x0_q4 = get_filter_offset(filter_x, filters_x);
const InterpKernel *const filters_y = get_filter_base(filter_y);
const int y0_q4 = get_filter_offset(filter_y, filters_y);
highbd_convolve(src, src_stride, dst, dst_stride,
filters_x, x0_q4, x_step_q4,
highbd_convolve(src, src_stride, dst, dst_stride, filters_x, x0_q4, x_step_q4,
filters_y, y0_q4, y_step_q4, w, h, bd);
}
void vpx_highbd_convolve8_avg_c(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, int bd) {
const int16_t *filter_y, int y_step_q4, int w,
int h, int bd) {
// Fixed size intermediate buffer places limits on parameters.
DECLARE_ALIGNED(16, uint16_t, temp[64 * 64]);
assert(w <= 64);
@@ -562,8 +546,8 @@ void vpx_highbd_convolve8_avg_c(const uint8_t *src, ptrdiff_t src_stride,
vpx_highbd_convolve8_c(src, src_stride, CONVERT_TO_BYTEPTR(temp), 64,
filter_x, x_step_q4, filter_y, y_step_q4, w, h, bd);
vpx_highbd_convolve_avg_c(CONVERT_TO_BYTEPTR(temp), 64, dst, dst_stride,
NULL, 0, NULL, 0, w, h, bd);
vpx_highbd_convolve_avg_c(CONVERT_TO_BYTEPTR(temp), 64, dst, dst_stride, NULL,
0, NULL, 0, w, h, bd);
}
void vpx_highbd_convolve_copy_c(const uint8_t *src8, ptrdiff_t src_stride,