Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(8)

Issues Search
    My Issues | Starred     Open | Closed | All

Unified Diff: source/libvpx/vp9/common/vp9_convolve.c

Issue 668403002: libvpx: Pull from upstream (Closed) Base URL: svn://svn.chromium.org/chrome/trunk/deps/third_party/libvpx/
Patch Set: Created 6 years, 2 months ago
Use n/p to move between diff chunks; N/P to move between comments. Draft comments are only viewable by you.
Jump to:
View side-by-side diff with in-line comments
Download patch
« no previous file with comments | « source/libvpx/vp9/common/vp9_convolve.h ('k') | source/libvpx/vp9/common/vp9_debugmodes.c » ('j') | no next file with comments »
Expand Comments ('e') | Collapse Comments ('c') | Show Comments Hide Comments ('s')
Index: source/libvpx/vp9/common/vp9_convolve.c
===================================================================
--- source/libvpx/vp9/common/vp9_convolve.c (revision 292608)
+++ source/libvpx/vp9/common/vp9_convolve.c (working copy)
@@ -130,7 +130,8 @@
// --Require an additional SUBPEL_TAPS rows for the 8-tap filter tails.
// --((64 - 1) * 32 + 15) >> 4 + 8 = 135.
uint8_t temp[135 * 64];
- int intermediate_height = (((h - 1) * y_step_q4 + 15) >> 4) + SUBPEL_TAPS;
+ int intermediate_height =
+ (((h - 1) * y_step_q4 + y0_q4) >> SUBPEL_BITS) + SUBPEL_TAPS;
assert(w <= 64);
assert(h <= 64);
@@ -137,9 +138,6 @@
assert(y_step_q4 <= 32);
assert(x_step_q4 <= 32);
- if (intermediate_height < h)
- intermediate_height = h;
-
convolve_horiz(src - src_stride * (SUBPEL_TAPS / 2 - 1), src_stride, temp, 64,
x_filters, x0_q4, x_step_q4, w, intermediate_height);
convolve_vert(temp + 64 * (SUBPEL_TAPS / 2 - 1), 64, dst, dst_stride,
@@ -284,11 +282,11 @@
}
#if CONFIG_VP9_HIGHBITDEPTH
-static void high_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) {
+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) {
int x, y;
uint16_t *src = CONVERT_TO_SHORTPTR(src8);
uint16_t *dst = CONVERT_TO_SHORTPTR(dst8);
@@ -301,7 +299,7 @@
int k, sum = 0;
for (k = 0; k < SUBPEL_TAPS; ++k)
sum += src_x[k] * x_filter[k];
- dst[x] = clip_pixel_high(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
+ dst[x] = clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
x_q4 += x_step_q4;
}
src += src_stride;
@@ -309,11 +307,11 @@
}
}
-static void high_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) {
+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) {
int x, y;
uint16_t *src = CONVERT_TO_SHORTPTR(src8);
uint16_t *dst = CONVERT_TO_SHORTPTR(dst8);
@@ -327,7 +325,7 @@
for (k = 0; k < SUBPEL_TAPS; ++k)
sum += src_x[k] * x_filter[k];
dst[x] = ROUND_POWER_OF_TWO(dst[x] +
- clip_pixel_high(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd), 1);
+ clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd), 1);
x_q4 += x_step_q4;
}
src += src_stride;
@@ -335,11 +333,11 @@
}
}
-static void high_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) {
+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) {
int x, y;
uint16_t *src = CONVERT_TO_SHORTPTR(src8);
uint16_t *dst = CONVERT_TO_SHORTPTR(dst8);
@@ -352,7 +350,7 @@
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_high(
+ dst[y * dst_stride] = clip_pixel_highbd(
ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
y_q4 += y_step_q4;
}
@@ -361,11 +359,11 @@
}
}
-static void high_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) {
+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) {
int x, y;
uint16_t *src = CONVERT_TO_SHORTPTR(src8);
uint16_t *dst = CONVERT_TO_SHORTPTR(dst8);
@@ -379,7 +377,7 @@
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_high(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd), 1);
+ clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd), 1);
y_q4 += y_step_q4;
}
++src;
@@ -387,13 +385,13 @@
}
}
-static void high_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) {
+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) {
// Note: Fixed size intermediate buffer, temp, places limits on parameters.
// 2d filtering proceeds in 2 steps:
// (1) Interpolate horizontally into an intermediate buffer, temp.
@@ -407,7 +405,8 @@
// --Require an additional SUBPEL_TAPS rows for the 8-tap filter tails.
// --((64 - 1) * 32 + 15) >> 4 + 8 = 135.
uint16_t temp[64 * 135];
- int intermediate_height = (((h - 1) * y_step_q4 + 15) >> 4) + SUBPEL_TAPS;
+ int intermediate_height =
+ (((h - 1) * y_step_q4 + y0_q4) >> SUBPEL_BITS) + SUBPEL_TAPS;
assert(w <= 64);
assert(h <= 64);
@@ -414,80 +413,77 @@
assert(y_step_q4 <= 32);
assert(x_step_q4 <= 32);
- if (intermediate_height < h)
- intermediate_height = h;
-
- high_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);
- high_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);
+ 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);
}
-void vp9_high_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) {
+void vp9_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 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;
- high_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 vp9_high_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, int bd) {
+void vp9_highbd_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, 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;
- high_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 vp9_high_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) {
+void vp9_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 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;
- high_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 vp9_high_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, int bd) {
+void vp9_highbd_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, 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;
- high_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 vp9_high_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) {
+void vp9_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 InterpKernel *const filters_x = get_filter_base(filter_x);
const int x0_q4 = get_filter_offset(filter_x, filters_x);
@@ -494,32 +490,32 @@
const InterpKernel *const filters_y = get_filter_base(filter_y);
const int y0_q4 = get_filter_offset(filter_y, filters_y);
- high_convolve(src, src_stride, dst, dst_stride,
- filters_x, x0_q4, x_step_q4,
- filters_y, y0_q4, y_step_q4, w, h, bd);
+ 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 vp9_high_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) {
+void vp9_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) {
// Fixed size intermediate buffer places limits on parameters.
DECLARE_ALIGNED_ARRAY(16, uint16_t, temp, 64 * 64);
assert(w <= 64);
assert(h <= 64);
- vp9_high_convolve8_c(src, src_stride, CONVERT_TO_BYTEPTR(temp), 64,
- filter_x, x_step_q4, filter_y, y_step_q4, w, h, bd);
- vp9_high_convolve_avg_c(CONVERT_TO_BYTEPTR(temp), 64, dst, dst_stride,
- NULL, 0, NULL, 0, w, h, bd);
+ vp9_highbd_convolve8_c(src, src_stride, CONVERT_TO_BYTEPTR(temp), 64,
+ filter_x, x_step_q4, filter_y, y_step_q4, w, h, bd);
+ vp9_highbd_convolve_avg_c(CONVERT_TO_BYTEPTR(temp), 64, dst, dst_stride,
+ NULL, 0, NULL, 0, w, h, bd);
}
-void vp9_high_convolve_copy_c(const uint8_t *src8, ptrdiff_t src_stride,
- uint8_t *dst8, 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 bd) {
+void vp9_highbd_convolve_copy_c(const uint8_t *src8, ptrdiff_t src_stride,
+ uint8_t *dst8, 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 bd) {
int r;
uint16_t *src = CONVERT_TO_SHORTPTR(src8);
uint16_t *dst = CONVERT_TO_SHORTPTR(dst8);
@@ -536,11 +532,11 @@
}
}
-void vp9_high_convolve_avg_c(const uint8_t *src8, ptrdiff_t src_stride,
- uint8_t *dst8, 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 bd) {
+void vp9_highbd_convolve_avg_c(const uint8_t *src8, ptrdiff_t src_stride,
+ uint8_t *dst8, 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 bd) {
int x, y;
uint16_t *src = CONVERT_TO_SHORTPTR(src8);
uint16_t *dst = CONVERT_TO_SHORTPTR(dst8);
« no previous file with comments | « source/libvpx/vp9/common/vp9_convolve.h ('k') | source/libvpx/vp9/common/vp9_debugmodes.c » ('j') | no next file with comments »

Powered by Google App Engine
This is Rietveld 408576698