Forum Discussion
FICO
3 years agoCommunity Manager
__lll_lock_wait allowing only one thread to work
I am calling the XPRESS optimizer from a multithreaded Pascal application and having performance issues with the solver.
While the program is multithreaded, all but one of the threads are sleeping all the time. A quick examination using gdb shows that those threads that are waiting are doing so after the call to XPRSdestroyprob:
(gdb) where
#0 0x00007ffff7bc175d in __lll_lock_wait () from /lib64/libpthread.so.0
#1 0x00007ffff7bbab44 in pthread_mutex_lock () from /lib64/libpthread.so.0
#2 0x00007ffff62981c1 in ?? () from /opt/xpressmp/lib/libxprs.so.41
#3 0x00007ffff4eb403e in XPRSdestroyprob_internal () from /opt/xpressmp/lib/libxprs.so.41
#4 0x00007ffff49881e9 in XPRSdestroyprob () from /opt/xpressmp/lib/libxprs.so.41
Am I calling the XPRESS solver library function in a wrong way?
Thank you very much.
21 Replies
- FICOCommunity Manager
Wow, thank you so much for taking the time to analyze each error in detail. I hope it can find a way to get a fix in a future version.
Thank you very much, Daniel!
Related Content
Related Content
Parameter Threads
Hi, I'm using Fico Xpress and I have a question about the parameter THREADS. According to the documentation, THREADS > 0 means the "number of threads to use". Let suppose THREADS = 4. In this ca...7 years agoPlace FICO® Xpress Optimization ForumFICO® Xpress Optimization Forum2 unread comments2 unread0Views0likes2CommentsConst correctness and thread safety in the C-API
How do I know : Which functions can be considered const i.e leave the XPRSprob in an unaltered state ? Which functions are thread safe ? (I am in particular interested in knowing if XPRSbtran and...3 years agoPlace FICO® Xpress Optimization ForumFICO® Xpress Optimization Forum3 unread comments3 unread0Views0likes3Comments