[Beowulf] progress on ifort and gfortran issue
Many of your questions may have already been answered in earlier discussions or in the FAQ. The search results page will indicate current discussions as well as past list serves, articles, and papers.
Nifty Tom Mitchell niftyompi at niftyegg.comTue Feb 24 11:00:16 PST 2009
- Previous message: [Beowulf] progress on ifort and gfortran issue
- Next message: [Beowulf] FW: Availability of an MPI program verification tool
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
On Tue, Feb 24, 2009 at 05:49:58PM -0000, Paulo Afonso Lopes wrote: > > If you compile your code with -g (debugging support), and "just run" > normally, it should crash and create a core file in the working dir. (do a > ls -la to find it) > > If you gdb --core=<your core file here> it will point you to the offending > instruction (helps if you are running on the dir where you have your > source code :-) otherwise you should add --directory=<your sources> > Do check user limits to ensure that the size of your core file is not limited to zero. On a cluster it is common to have the core file limit set to zero. Should a many rank MPI program dump core the I/O tossed at a NFS server can boggle the mind. The value of -g is that symbols are kept and associated with bits in the binary so a debugger can interact symbolical with the binary and source code symbols. $ ulimit -a core file size (blocks, -c) 0 data seg size (kbytes, -d) unlimited scheduling priority (-e) 0 file size (blocks, -f) unlimited .... virtual memory (kbytes, -v) unlimited file locks (-x) unlimited Also the name of core files can be manipulated so each rank has its own core file. -- T o m M i t c h e l l Found me a new hat, now what?
- Previous message: [Beowulf] progress on ifort and gfortran issue
- Next message: [Beowulf] FW: Availability of an MPI program verification tool
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
