From: Maxime Boissonneault (maxime.boissonneault_at_calculquebec.ca)
Date: Thu Jul 27 2017 - 13:47:54 CDT
Hi Mauro,
We are already building from source. MPI will work, but that will not
support GPUs. We need GPUs + OPA, and the verbs version does not seem to
want to work.
Maxime
On 17-07-27 14:06, Mauro Lapelosa wrote:
> Hi Maxime,
> You may want to compile the source.
>
> ./build charm++ mpi-linux-x86_64 smp --with-production -j8
>
> and for NAMD is something like:
>
> ./config Linux-x86_64-g++ --charm-arch mpi-linux-x86_64
>
> It should work.
>
> Best Regards,
>
> Mauro Lapelosa
>
> On Thu, Jul 27, 2017 at 8:00 PM, Maxime Boissonneault
> <maxime.boissonneault_at_calculquebec.ca
> <mailto:maxime.boissonneault_at_calculquebec.ca>> wrote:
>
> Hi,
>
> We just deployed a cluster with Intel OmniPath instead of
> InfiniBand as a network fabric. We are having problems running
> NAMD on that hardware. It works within a node, but not between nodes.
>
> Is there anything special we should do for compiling it to work on
> this ?
>
>
> Thanks,
>
> --
> ---------------------------------
> Maxime Boissonneault
> Analyste de calcul - Calcul Québec, Université Laval
> Président - Comité de coordination du soutien à la recherche de
> Calcul Québec
> Team lead - Research Support National Team, Compute Canada
> Instructeur Software Carpentry
> Ph. D. en physique
>
>
-- --------------------------------- Maxime Boissonneault Analyste de calcul - Calcul Québec, Université Laval Président - Comité de coordination du soutien à la recherche de Calcul Québec Team lead - Research Support National Team, Compute Canada Instructeur Software Carpentry Ph. D. en physique
This archive was generated by hypermail 2.1.6 : Mon Dec 31 2018 - 23:20:28 CST