- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi All,
I'm having a problem with ifort. If I compile my code using...
ifort -D_LINUX -static -openmp -funroll-loops -parallel -ipo -O3 -axS x3.f -o x3_42_omp
I have no problems running it.
If I add the -r8 flag...
ifort -r8 -D_LINUX -static -openmp -funroll-loops -parallel -ipo -O3 -axS x3.f -o x3_42_omp
I get a seg fault.
I'm using...
$ ifort -v
Version 11.1
on...
$ uname -a
Linux cs04r-sc-com01-40.diamond.ac.uk 2.6.18-92.1.10.el5_lustre.1.6.6smp #1 SMP Tue Aug 26 12:16:17 EDT 2008 x86_64 x86_64 x86_64 GNU/Linux
Has anyone seen this kind of error before? is there a workaround? Am I doing something stupid???
Thanks for any help,
Wes.
I'm having a problem with ifort. If I compile my code using...
ifort -D_LINUX -static -openmp -funroll-loops -parallel -ipo -O3 -axS x3.f -o x3_42_omp
I have no problems running it.
If I add the -r8 flag...
ifort -r8 -D_LINUX -static -openmp -funroll-loops -parallel -ipo -O3 -axS x3.f -o x3_42_omp
I get a seg fault.
I'm using...
$ ifort -v
Version 11.1
on...
$ uname -a
Linux cs04r-sc-com01-40.diamond.ac.uk 2.6.18-92.1.10.el5_lustre.1.6.6smp #1 SMP Tue Aug 26 12:16:17 EDT 2008 x86_64 x86_64 x86_64 GNU/Linux
Has anyone seen this kind of error before? is there a workaround? Am I doing something stupid???
Thanks for any help,
Wes.
Link Copied
2 Replies
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
If there is an array out-of-bounds problem, it could get much worse with -r8. Or, you may have something in your source which isn't consistent with -r8. Among many possibilities, remember that real*4 is not affected by -r8.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Quoting - tim18
If there is an array out-of-bounds problem, it could get much worse with -r8. Or, you may have something in your source which isn't consistent with -r8. Among many possibilities, remember that real*4 is not affected by -r8.
Hi Tim,
Thanks for the response.
I've managed to narrow the problem down. If I use -r8 -parallel the code seg faults. If I just use -parallel I have no problems.
A friend suggested this might be due to Rhel5 using randomised memory mapping. Certainly it wasn't an issue running on a rhel4 kernel.
Thanks again,
Wes.
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