liz/rawTherapee
liz/rawTherapee
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Capture sharpening: add vectorization hints for gcc

Browse Source
This commit is contained in:
Ingo Weyrich 2020-01-10 19:45:59 +01:00
parent 355fd5a44a
commit 09b1a5fada
1 changed files with 18 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
rtengine/capturesharpening.cc
View File

@ -131,8 +131,10 @@ inline void gauss3x3div (float** RESTRICT src, float** RESTRICT dst, float** RES
const float c00 = kernel[1][1];
for (int i = 1; i < tileSize - 1; i++) {
#ifdef __clang__
#if defined(__clang__)
#pragma clang loop vectorize(assume_safety)
#elif defined(__GNUC__)
#pragma GCC ivdep
#endif
for (int j = 1; j < tileSize - 1; j++) {
const float val = c11 * (src[i - 1][j - 1] + src[i - 1][j + 1] + src[i + 1][j - 1] + src[i + 1][j + 1]) +
@ -154,8 +156,10 @@ inline void gauss5x5div (float** RESTRICT src, float** RESTRICT dst, float** RES
for (int i = 2; i < tileSize - 2; ++i) {
// I tried hand written SSE code but gcc vectorizes better
#ifdef __clang__
#if defined(__clang__)
#pragma clang loop vectorize(assume_safety)
#elif defined(__GNUC__)
#pragma GCC ivdep
#endif
for (int j = 2; j < tileSize - 2; ++j) {
const float val = c21 * ((src[i - 2][j - 1] + src[i - 2][j + 1]) + (src[i - 1][j - 2] + src[i - 1][j + 2]) + (src[i + 1][j - 2] + src[i + 1][j + 2]) + (src[i + 2][j - 1] + src[i + 2][j + 1])) +
@ -183,8 +187,10 @@ inline void gauss7x7div(float** RESTRICT src, float** RESTRICT dst, float** REST
for (int i = 3; i < tileSize - 3; ++i) {
// I tried hand written SSE code but gcc vectorizes better
#ifdef __clang__
#if defined(__clang__)
#pragma clang loop vectorize(assume_safety)
#elif defined(__GNUC__)
#pragma GCC ivdep
#endif
for (int j = 3; j < tileSize - 3; ++j) {
const float val = c31 * ((src[i - 3][j - 1] + src[i - 3][j + 1]) + (src[i - 1][j - 3] + src[i - 1][j + 3]) + (src[i + 1][j - 3] + src[i + 1][j + 3]) + (src[i + 3][j - 1] + src[i + 3][j + 1])) +
@ -208,8 +214,10 @@ inline void gauss3x3mult(float** RESTRICT src, float** RESTRICT dst, const int t
const float c00 = kernel[1][1];
for (int i = 1; i < tileSize - 1; i++) {
#ifdef __clang__
#if defined(__clang__)
#pragma clang loop vectorize(assume_safety)
#elif defined(__GNUC__)
#pragma GCC ivdep
#endif
for (int j = 1; j < tileSize - 1; j++) {
const float val = c11 * (src[i - 1][j - 1] + src[i - 1][j + 1] + src[i + 1][j - 1] + src[i + 1][j + 1]) +
@ -232,8 +240,10 @@ inline void gauss5x5mult (float** RESTRICT src, float** RESTRICT dst, const int
for (int i = 2; i < tileSize - 2; ++i) {
// I tried hand written SSE code but gcc vectorizes better
#ifdef __clang__
#if defined(__clang__)
#pragma clang loop vectorize(assume_safety)
#elif defined(__GNUC__)
#pragma GCC ivdep
#endif
for (int j = 2; j < tileSize - 2; ++j) {
const float val = c21 * ((src[i - 2][j - 1] + src[i - 2][j + 1]) + (src[i - 1][j - 2] + src[i - 1][j + 2]) + (src[i + 1][j - 2] + src[i + 1][j + 2]) + (src[i + 2][j - 1] + src[i + 2][j + 1])) +
@ -261,8 +271,10 @@ inline void gauss7x7mult(float** RESTRICT src, float** RESTRICT dst, const int t
for (int i = 3; i < tileSize - 3; ++i) {
// I tried hand written SSE code but gcc vectorizes better
#ifdef __clang__
#if defined(__clang__)
#pragma clang loop vectorize(assume_safety)
#elif defined(__GNUC__)
#pragma GCC ivdep
#endif
for (int j = 3; j < tileSize - 3; ++j) {
const float val = c31 * ((src[i - 3][j - 1] + src[i - 3][j + 1]) + (src[i - 1][j - 3] + src[i - 1][j + 3]) + (src[i + 1][j - 3] + src[i + 1][j + 3]) + (src[i + 3][j - 1] + src[i + 3][j + 1])) +