- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hello,
I'm trying to compile and run the full SPEC CPU2006 benchmark suite using the Intel C++ and Fortran 9.1 compilers. Works fine, except for one benchmark: 410.bwaves (CFP2006, see http://www.spec.org/cpu2006/Docs/410.bwaves.html for more info).
My platform is IA-32 (Intel Pentium4) running Fedora Linux Core 4.
This Fortran benchmark compiles without problems, using the '-fast' optimization flag, together with the -prof-gen and -prof-use flags (thus using 2 passes).
Running the benchmark results in a segmentation fault, which doesn't occur when compiling with gcc-4.1.1 -O2. Using '-O2' instead of -'fast' (with ifort) didn't help, removing the -prof-gen en -prof-use flags didn't help either.
I've tried running idb on it, but since I don't know any Fortran that didn't help me at all.
Is this a known problem, and is there any solution for it? I've searched this forum (without luck), and neither the SPEC website or Google provide me with much information.
I'll be happy to provide more information if needed.
Thanks,
Kenneth
I'm trying to compile and run the full SPEC CPU2006 benchmark suite using the Intel C++ and Fortran 9.1 compilers. Works fine, except for one benchmark: 410.bwaves (CFP2006, see http://www.spec.org/cpu2006/Docs/410.bwaves.html for more info).
My platform is IA-32 (Intel Pentium4) running Fedora Linux Core 4.
This Fortran benchmark compiles without problems, using the '-fast' optimization flag, together with the -prof-gen and -prof-use flags (thus using 2 passes).
Running the benchmark results in a segmentation fault, which doesn't occur when compiling with gcc-4.1.1 -O2. Using '-O2' instead of -'fast' (with ifort) didn't help, removing the -prof-gen en -prof-use flags didn't help either.
I've tried running idb on it, but since I don't know any Fortran that didn't help me at all.
Is this a known problem, and is there any solution for it? I've searched this forum (without luck), and neither the SPEC website or Google provide me with much information.
I'll be happy to provide more information if needed.
Thanks,
Kenneth
Link Copied
5 Replies
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Which specific Fortran compiler are you using> 9.1.??? I am not familiar with the details of using the Intel compilers with SPEC, but if you are seeing a problem using a current compiler (9.1.037 I think is latest?), then please contact Intel Premier Support for help.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I'm using the 9.1.036 Intel Fortran compiler, which I download last weekend from the website. I'll try the 037 version which is now online I see, but I don't think I can contact Premium Support (I'm using a non-commercial license)... Any other ways if 037 doesn't work for me?
edit:
ifort 9.1.037 doesn't seem to work either.
Sample of the output (generated by runspec):
Compiling Binaries
Building 410.bwaves ref base kehoste_linux_ia32_ic9.1 default: (build_base_kehoste_linux_ia32_ic9.1.0000)
Build successes: 410.bwaves(base)
Running Benchmarks
Running (#1) 410.bwaves ref base kehoste_linux_ia32_ic9.1 default
410.bwaves: copy #0 non-zero return code (rc=0, signal=11)
Running bwaves without runspec show a Segmentation fault almost immediately.
Any pointers are welcome, since I can't use the Intel support.
edit:
ifort 9.1.037 doesn't seem to work either.
Sample of the output (generated by runspec):
Compiling Binaries
Building 410.bwaves ref base kehoste_linux_ia32_ic9.1 default: (build_base_kehoste_linux_ia32_ic9.1.0000)
Build successes: 410.bwaves(base)
Running Benchmarks
Running (#1) 410.bwaves ref base kehoste_linux_ia32_ic9.1 default
410.bwaves: copy #0 non-zero return code (rc=0, signal=11)
Running bwaves without runspec show a Segmentation fault almost immediately.
Any pointers are welcome, since I can't use the Intel support.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Yes, you can use Intel Premier Support to submit an issue. Responses aren't guaranteed and you're not eligible for updates, but we'll do the best we can. You need to have registered for support - if you didn't do that, go back to the Intel Registration Center and re-register your serial number, being sure to check the box for support registration.
We have lots of SPEC experts (I'm not one of them).
We have lots of SPEC experts (I'm not one of them).
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Thank you for your suggestion. I've reported my problem to Premium Support, and I hope they will be able to help me. For reference: Premium Support issue number 394536.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Thanks to the people from Premium Support, we were able to pin down the problem to stack size. Apparently, ifort stresses the stack more than gcc does, which was the cause of this problem... Changing "ulimit -s" solved it.

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