We have been chasing this problem forever and hope to get some direction for this problem. We use Intel MKL for one of our internal FORTRAN programs and recently started to use MKL-Parallel option (/Qmkl:parallel) and have been experiencing problems:
Sometimes, the program seems to crash / hangs executing the math functions and sometimes it doesn't come back as if it is silently crashing.
- How can we gracefully capture the fatal errors from the calls to MKL Libraries?
- How do we handle the cases where the threads do not exit?
Here's a case below where the MKL thread doesn't exit.
We limited the OMP threads to two and set the stack size to 200000000. For this specific input, it hangs in there as shown below ( from the process explorer).
Private bytes committed : 1.2GB. Is the stack size too big??
Could you please tell some background formation:
1. what compiler is you using for compiling the r_frame.exe?
2. are you worked on 32bit or 64bit application, do you have other thread method in r_frame.exe? if use and MKL-Sequential option (/Qmkl:sequential), do the application work?
3. What MKL functions are used? and what function may cause the problem?
Compiler used is: Intel(R) Visual Fortran Compiler XE 188.8.131.52 [IA-32]...Win32
DGEMM for matrix multiplication has been extensively used and this seems to be the last function that got called before it hung. However, the program seems to be completely run since we got the reports okay, but it doesn;t exit. For some other jobs, the behavior is different, it hangs in the middle without reaching to the end.
Sequential run is working just fine.
Thanks for looking into this.
So it looks unknown issue.
1. Could you please try the latest Intel Fortran 15.0.2 ?
Then you will get email with download link and MKL should provide verbose in the new version. Please see
So if you switch on the option, you will see MKL functions status.
Moreever,have you tried some debug tool or threading tools to check the problem to see where is the exact problem.
like Intel® Inspector XE Memory & threading dynamic analysis for code quality?