[bolt-discuss] Issue with Bolt + FORTRAN?

[Halim Amer]

Hi Aidan,

Thanks for reporting this problem. You can track the progress on this 
issue at https://github.com/pmodels/bolt/issues/4. We are currently 
constrained by some deadlines, but we should be able to work on this 
issue in the upcoming few weeks.

Halim
www.mcs.anl.gov/~aamer

On 3/22/17 9:48 AM, aidan.chalk at stfc.ac.uk wrote:
> Hi,
>
> I'm trying to use Bolt to compare performance with Intel/gcc openMP runtimes with taskyield. The code in question is a FORTRAN code, however I'm having issues using bolt (including make ftest)
>
> I'm compiling with Intel 17.2.050, and have attached the c.txt, m.txt and mi.txt discussed in your build process.
>
> To run the make ftest I removed the enforcing of -lm (because this is unneeded with intel compiler) and:
> bash-4.1$ export TEST_FC=ifort
> bash-4.1$ export TEST_FFLAGS="-g -O2 -qopenmp -I/~/INSTALL/bolt/include -L/~/INSTALL/bolt/lib -Wl,-rpath=~/INSTALL/bolt/lib"
>
> when  the tests progress a little, I get this error:
> Testing for "omp_threadprivate":
> Generating sources .............. success
> Compiling soures ................ success
> Running test with 8 threads ../bin/fortran/test_omp_threadprivate: relocation error: ./bin/fortran/test_omp_threadprivate: symbol kmp_aligned_malloc, version VERSION not defined in file libiomp5.so with link time reference
> .... failed 100% of the tests
>
>
> The code I which I want to use gives me the same error (and also uses threadprivate variables. Is this something that isn't supported with FORTRAN yet?
>
> Also  do_collapse causes an internal compiler error with Intel 17.2.050
> Testing for "do_collapse":
> Generating sources .............. success
> Compiling soures ............bin/fortran/test_do_collapse.f(46): internal error: Please visit 'http://www.intel.com/software/products/support' for assistance.
> !$omp end do
> ^
> [ Aborting due to internal error. ]
> compilation aborted for bin/fortran/test_do_collapse.f (code 1)
>
>
> I already ran the ctests and these worked correctly (a few tests failed but not in such a way)
>
> Thanks,
> Aidan Chalk
>
>
> High Performance Software Engineer
> Hartree Centre
>
>
>
>
> _______________________________________________
> discuss mailing list
> discuss at lists.bolt-omp.org
> https://lists.bolt-omp.org/mailman/listinfo/discuss
>