- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
the following codes
"real*8 ,allocatable::a(:,:,:)
allocate(a(3000,100,100))
a=-10"
will spend too many time,
i want to parallel to speed, how to do?
Link Copied
1 Reply
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
compile option -Qparallel ought to do as well as anything, in such a simple case. For OpenMP ( compile switch /Qopenmp), you would write
....
!$omp parallel do
do i=1,size(a,3)
!dir$ vector nontemporal
a(:,:,i) = -10
end do
Significant speedup on a dual CPU platform is likely to depend on having NUMA BIOS option set. There may not be much speedup on a single dual core CPU, none on single core HT.
....
!$omp parallel do
do i=1,size(a,3)
!dir$ vector nontemporal
a(:,:,i) = -10
end do
Significant speedup on a dual CPU platform is likely to depend on having NUMA BIOS option set. There may not be much speedup on a single dual core CPU, none on single core HT.
Reply
Topic Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page