portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-10-30 19:44:32 +00:00
Code Releases Activity

Faster gather

Browse Source
This commit is contained in:
paboyle
2017-02-16 23:51:15 +00:00
parent bd600702cf
commit be3a8249c6
1 changed files with 5 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
lib/Stencil.h
View File

@@ -240,6 +240,10 @@ PARALLEL_FOR_LOOP
for(int o=0;o<Mergers[i].buffer_size/2;o++){ for(int o=0;o<Mergers[i].buffer_size/2;o++){
exchange(Mergers[i].mpointer[2*o],Mergers[i].mpointer[2*o+1], exchange(Mergers[i].mpointer[2*o],Mergers[i].mpointer[2*o+1],
Mergers[i].vpointers[0][o],Mergers[i].vpointers[1][o],Mergers[i].type); Mergers[i].vpointers[0][o],Mergers[i].vpointers[1][o],Mergers[i].type);
// cobj temp1,temp2;
// exchange(temp1,temp2,Mergers[i].vpointers[0][o],Mergers[i].vpointers[1][o],Mergers[i].type);
// vstream(Mergers[i].mpointer[2*o],temp1);
// vstream(Mergers[i].mpointer[2*o+1],temp2);
} }
} }
mergetime+=usecond(); mergetime+=usecond();
@@ -1037,9 +1041,7 @@ PARALLEL_FOR_LOOP
int sx = (x+sshift)%rd; int sx = (x+sshift)%rd;
gathermtime-=usecond(); gathermtime+=Gather_plane_exchange(rhs,spointers,dimension,sx,cbmask,compress,permute_type);
Gather_plane_exchange(rhs,spointers,dimension,sx,cbmask,compress,permute_type);
gathermtime+=usecond();
//spointers[0] -- low //spointers[0] -- low
//spointers[1] -- high //spointers[1] -- high