You are here

program not terminating...

3 posts / 0 new
Last post
program not terminating...
#1

I built a 64-bit version of rosetta (2.3.0) using Intel v.91 compilers with MPICH 1.2.7p1 on a 64-bit AMD processor. I compiled the code with a symbol table and no optimization and ran it in the totalview debugger with memory debugging turned on. At the very end of the program when doing memory cleanup (job_distributor::memory_cleanup), there is a memory error when deallocating memory. The error is "program attempted to free a block incorrectly in the middle of the block." The location of the error is line 768 in file FArrayB.hh

This causes the program to hang and never complete.

How do I fix this?

Thank you
Andrew

Wed, 2009-02-25 06:40
atoler

I do not release understand your question. Did the error happen during the compiling or simulation runs?
If it happened during a simulation run, can you provide the command line to us?

> I built a 64-bit version of rosetta (2.3.0) using Intel v.91 compilers with MPICH 1.2.7p1 on a 64-bit AMD processor. I compiled the code with a symbol table and no optimization and ran it in the totalview debugger with memory debugging turned on. At the very end of the program when doing memory cleanup (job_distributor::memory_cleanup), there is a memory error when deallocating memory. The error is "program attempted to free a block incorrectly in the middle of the block." The location of the error is line 768 in file FArrayB.hh
>
> This causes the program to hang and never complete.
>
> How do I fix this?
>
> Thank you
> Andrew

Wed, 2009-02-25 08:40
yiliu

Rosetta compiles.

The behavior is when the application runs - it will not terminate. The process must be killed if run interactively, or reach the end of walltime if I submit as a batch job, (which makes it impossible for me to estimate walltimes)

Here's one that produces output, but the program will not terminate after creating the result (1000 .pdb's)

./rosetta aa prot1 _ -nstruct 1000 -paths /paths.txt

The predictions (runpsipred_single) and fragments (make_fragments.pl) have been run previously and are there for rosetta when I run it - I don't get errors that pieces and parts are missing. It just won't terminate when it seems to be done processing. If I turn off -silent, as in the example, rosetta will output the 1000 .pdb files, but not terminate after the last one is written. Just sits there, cranking 100% cpu.

Wed, 2009-02-25 17:23
atoler