ftinetti wrote:Hi Wenby,
Please take a look and try setting OMP_STACKSIZE according to your needs. I don't know how the stack is handled in windows ifort, though... there are several issues in Linux/UNIX, but usually eventual problems are solved playing around ulimit...
Yes, I know this solution in Linux. But I wanna to find such kind solution in Windows.....
ftinetti wrote:Yes, I know this solution in Linux. But I wanna to find such kind solution in Windows.....
I see, did you try setting OMP_STACKSIZE?
A quick search provided
http://software.intel.com/en-us/article ... ion-fault/
Did you see/use it?
I have already tried /Fn compiler option, it looks like can work. but when I excute the .exe file after compiled, Windows always showed "*.exe has stop working, windows can check this problem......"
I use the simplyest OpenMp test code "hello world", the result is also same!
ftinetti wrote:I use the simplyest OpenMp test code "hello world", the result is also same!
Strange... would you post the code? maybe the one you were referring to
ftinetti wrote:I see...
1) Include the declarations of every variable (not those in the subprograms, just the ones in the code you send). Furthermore, it would be very useful if you can post a complete example based on this code, i.e. one which could be compiled and run. Leave subprograms without code. Later, you could add the local data and later, the code.
2) Take into account that subprograms should be thread-safe. Common blocks, I/O, and save variables usually imply that subprograms are not thread safe.
Users browsing this forum: Google [Bot] and 10 guests