This may not be a proper question to ask here, however I don't find a better place to ask.
I am testing Ubuntu Natty 11.04. I build R with Blas/Lapack linked to intel MKL. Everything works fine but when quitting R with q(), the R process won't quit, and the system monitor shows its waiting channel as futex_wait_queue_me. For example, with "make check", the process will halted after checking the graphics package, whose test script called lm(), which I think used lapack routines in MKL. The next thing I tried is using environmental variables such that MKL runs in sequential mode. Then everything goes perfect. "make check" was successful. So I identified that the problem is with the multi-threaded version of MKL. After some trials, I think it is true, as eigen(), inv(), and matrix multiplication, etc all cause the same problem. The computing itself is fine, but R is halted when quiting. Then I wrote some C code using blas/lapack to test, and there's no problem with MKL, at least with my testing code. Both R-2.13 and R-2.12 are tried. Both intel icc and gcc, ifort and gfortran are tried. In sum, I tried most configuration combination of compilers, openmp flags, etc I can think about. But the problem persist. I think this could be a problem with ubuntu 11.04. So this as I said in the beginning, this may not be a proper question to ask here. However, I tried blas/lapack routines for decomposition, etc in C, and it runs without problem. So there's may be some workaround with the problem when building R and hopefully someone can give some hints. Best, Yan ______________________________________________ [email protected] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
