Skip to content

Vectorize inner loop of get_tiles and out_transform with SIMD #3

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
yolanda15 wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Vectorize inner loop of get_tiles and out_transform with SIMD #3

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 4 additions & 7 deletions src/falcon.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -56,8 +56,8 @@ static void get_tiles(const float* restrict image, const int ldi, const int irow

// work on one image plane at a time, irrespective of the order
for(i = 0; i < irows-2; i += 2){
#pragma unroll(4)
for(j = 0; j < (irows-2); j += 2){
#pragma omp simd linear(j:2, tile_count:1)
for(j = 0; j < (irows-2); j += 2, tile_count++){
tmp[0 :4] =data[(i+0)*ldi+j:4];
tmp[4 :4] =data[(i+1)*ldi+j:4];
tmp[8 :4] =data[(i+2)*ldi+j:4];
Expand Down Expand Up @@ -98,9 +98,6 @@ static void get_tiles(const float* restrict image, const int ldi, const int irow
otile[tile_count+13*STRIDE] = s[13];
otile[tile_count+14*STRIDE] = s[14];
otile[tile_count+15*STRIDE] = s[15];


tile_count++;
}
}
}
Expand Down Expand Up @@ -219,7 +216,8 @@ static void out_transform(const float* restrict d, const int K, const int ntiles
int i, j;
// work on one output plane at a time, irrespective of the order
for(i = 0; i < oH; i += 2){
for(j = 0; j < oW; j += 2){
#pragma omp simd linear(j:2, tile_offset:1)
for(j = 0; j < oW; j += 2, tile_offset++){

// gather the 16 elements form C to form a tile
c1[0 ] = d[tile_offset+0 *STRIDE];
Expand Down Expand Up @@ -258,7 +256,6 @@ static void out_transform(const float* restrict d, const int K, const int ntiles
data[i*ldo+j+1] =c2[1];
data[(i+1)*ldo+j] = c2[2];
data[(i+1)*ldo+j+1] = c2[3];
tile_offset++;
}
}
}
Expand Down