Newsgroups: comp.programming.threads,comp.parallel.mpi From: Alexandre Carissimi Subject: Re: BUG on Threads IRIX6.4?? Organization: LMC-Laboratoire de Modelisation et Calcul Date: Mon, 23 Feb 1998 13:53:44 +0100 Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Message-ID: <34F17158.6D5E@imag.fr> Eric Salo wrote: > details of our thread library to answer your > question, but I suspect that your problem is > related to the fact that in our MPI 3.0 release, > the pid which calls MPI_Init() never returns to > the application; it just forks off the compute > processes and then stays in a select() loop until > the end of the job. > > Assuming that I am correct, you should have better > luck with our new MPI 3.1 release, which instead > creates all of the compute processes before the > application ever reaches main(). Eric, Thanks for your answer! But, unfortunately changing MPI doesn't resolve the problem. My version now is SGI MPI 3.1.x.8 (obtained with -v option when I execute it). After my first posting I continued to do some tests and I observed that (complementar infos): - IF I CHANGE threads priorities and I use mpirun to execute my test program, the main thread id is changed. For example, main thread id is 10000 and after priority changing it becomes 30000. If I create some others threads, their priorities are allocate as 30001, 30002, etc... NO MPI CALL IS DONE!! I use only mpirun to execute the application. -Using EXACTLY the same program test, but now, I execute WITHOUT MPIRUN, the results are: main thread id=10000, after priority change it continues as 10000, if I create other threads their ID are 100001, 100002. So, it seems pthreads library works well There is something strange on mpirun command. I would like to know where I can fin more technical informations about the threads implementation (SGI) and about initialization steps on MPI. These kind of infos are available?? I hope find some way to avoid this problem because threads ID are importante for us. Thanks. __________________________________________________________________ CARISSIMI, Alexandre Alexandre.Carissimi@imag.fr Laboratoire de Modelisation et Calcul - LMC Institut National Polytechnique de Grenoble - INPG 100, Rue des Mathematiques Fone: (33) 04.76.51.46.03 38041 - Grenoble - Cedex 9 FRANCE Fax : (33) 04.76.51.48.46 (International calls removes 0 from 04) "Le seul moteur pour gagner c'est innover" Renault (F1 - 1996) ________________________________________________________________