[petsc-dev] Fwd: hbw_malloc on KNL

[Barry Smith]

> Begin forwarded message:
> 
> From: Hong Zhang <hongzhang at anl.gov>
> Subject: hbw_malloc on KNL
> Date: August 31, 2016 at 5:00:44 PM CDT
> To: "Smith, Barry F." <bsmith at mcs.anl.gov>, "Balay, Satish" <balay at mcs.anl.gov>, "Schanen, Michel" <mschanen at anl.gov>
> 
> Hi all,
> 
> I am experimenting with hbw_malloc() in PETSc. I used it to replace the default malloc() in PETSc. But I got the following error messages on the screen when running my PETSc example:
> 
> map size mismatch; abort
> : No such file or directory
> 
> Each process threw out the above message twice. After that, the program continued to run to the end. But it took way more time than expected.
> 
> The good news is, if I call hbw_malloc() to allocate a bit memory, e.g. 1 byte, before PetscInitialize(), the error will disappear and the example runs almost as fast as using the numactl option (which automatically put everything on MCDRAM). Although it works somehow, adding a test call to how_malloc() like this is just absurd. I am wondering if you have any clue on why this happens and how to fix it. FYI, I attached the driver for the PETSc example.
> 
> Thanks,
> Hong
>  

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20160831/f510e0af/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ex5_knl.c
Type: application/octet-stream
Size: 15252 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20160831/f510e0af/attachment.obj>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20160831/f510e0af/attachment-0001.html>