ref: 23d1c06268e884ee65d3d67943c62c26a7e369b7
parent: 87b2495f952b86709c673d923d6f60dfe5df852d
author: Scott LaVarnway <[email protected]>
date: Mon Oct 5 02:33:42 EDT 2015
VPX: refactor vpx_idct32x32_1_add_sse2() Change-Id: Ia1a2cac0e9dc05f3207b3433a6c1589fa7f2aee3
--- a/vpx_dsp/x86/inv_txfm_sse2.c
+++ b/vpx_dsp/x86/inv_txfm_sse2.c
@@ -3474,7 +3474,7 @@
int stride) {
__m128i dc_value;
const __m128i zero = _mm_setzero_si128();
- int a, i;
+ int a, j;
a = dct_const_round_shift(input[0] * cospi_16_64);
a = dct_const_round_shift(a * cospi_16_64);
@@ -3482,12 +3482,11 @@
dc_value = _mm_set1_epi16(a);
- for (i = 0; i < 4; ++i) {
- int j;
- for (j = 0; j < 32; ++j) {
- RECON_AND_STORE(dest + j * stride, dc_value);
- }
- dest += 8;
+ for (j = 0; j < 32; ++j) {
+ RECON_AND_STORE(dest + 0 + j * stride, dc_value);
+ RECON_AND_STORE(dest + 8 + j * stride, dc_value);
+ RECON_AND_STORE(dest + 16 + j * stride, dc_value);
+ RECON_AND_STORE(dest + 24 + j * stride, dc_value);
}
}