[CP2K-user] [CP2K:12845] 64 water benchmark slow with multiple nodes

Krack Matthias (PSI) matthi... at psi.ch
Wed Feb 5 09:00:58 UTC 2020


Hi Stacey

With that limited information one can only guess the cause of that behavior. I would check firstly, if the right device for the internode communication is used by the MPI installation CP2K is built with. Your admins may help you to figure that out.

Best

Matthias

-----Original Message-----
From: cp... at googlegroups.com <cp... at googlegroups.com> On Behalf Of MD Simulation
Sent: Mittwoch, 5. Februar 2020 03:07
To: cp2k <cp... at googlegroups.com>
Subject: [CP2K:12845] 64 water benchmark slow with multiple nodes

Hello,

I am running the 64 water benchmark and I see a huge slow down when I use more than one node:

1 node, 24 cores, 55s
2 nodes, 48 cores, 700s

I suspect I've done something wrong during compilation. Any ideas what would cause the slow down? 

Thanks 
Stacey 

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp... at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/ecc57f76-214d-4e53-b477-1e03eaac09ec%40googlegroups.com.


More information about the CP2K-user mailing list