ref: 47174d60c86e6092ce75cdcf97df555762876389
parent: 146005a9114278288ca77d2712c4c79e7026eecb
parent: d204c4bf017dc8313fc315f5c4da4492acd7641f
author: Johann Koenig <johannkoenig@google.com>
date: Fri May 26 13:25:23 EDT 2017
Merge "Use vdup instead of vmov"
--- a/vpx_dsp/arm/subpel_variance_neon.c
+++ b/vpx_dsp/arm/subpel_variance_neon.c
@@ -30,8 +30,8 @@
int pixel_step,
unsigned int output_height,
const uint8_t *filter) {
- const uint8x8_t f0 = vmov_n_u8(filter[0]);
- const uint8x8_t f1 = vmov_n_u8(filter[1]);
+ const uint8x8_t f0 = vdup_n_u8(filter[0]);
+ const uint8x8_t f1 = vdup_n_u8(filter[1]);
unsigned int i;
for (i = 0; i < output_height; i += 2) {
const uint8x8_t src_0 = load_unaligned_u8(src_ptr, src_pixels_per_line);
@@ -54,8 +54,8 @@
int pixel_step,
unsigned int output_height,
const uint8_t *filter) {
- const uint8x8_t f0 = vmov_n_u8(filter[0]);
- const uint8x8_t f1 = vmov_n_u8(filter[1]);
+ const uint8x8_t f0 = vdup_n_u8(filter[0]);
+ const uint8x8_t f1 = vdup_n_u8(filter[1]);
unsigned int i;
for (i = 0; i < output_height; ++i) {
const uint8x8_t src_0 = vld1_u8(&src_ptr[0]);
@@ -78,8 +78,8 @@
unsigned int output_height,
unsigned int output_width,
const uint8_t *filter) {
- const uint8x8_t f0 = vmov_n_u8(filter[0]);
- const uint8x8_t f1 = vmov_n_u8(filter[1]);
+ const uint8x8_t f0 = vdup_n_u8(filter[0]);
+ const uint8x8_t f1 = vdup_n_u8(filter[1]);
unsigned int i, j;
for (i = 0; i < output_height; ++i) {
for (j = 0; j < output_width; j += 16) {