[petsc-dev] MKL memory leaks

Jed Brown jedbrown at mcs.anl.gov
Wed Nov 6 07:34:21 CST 2013

Stefano Zampini <s.zampini at cineca.it> writes:

> I noticed that recent versions of MKL produce memory leaks if the memory
> management functions are not properly called at the end of the program.
> http://software.intel.com/en-us/node/438664
> Do we need to add those calls in PetscFinalize or it is user responsibility?

Hmm, looks like the user could run with MKL_DISABLE_FAST_MM=1.  Maybe we
should call mkl_disable_fast_mm() when running in debug mode.  I don't
know if calling mkl_free_buffers has an impact on other running parts of
a program (if the user is calling MKL independent of PETSc).
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 835 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20131106/9f1fa89c/attachment.sig>

More information about the petsc-dev mailing list