[Users] [ET Trac] [Einstein Toolkit] #2008: NaNs when running static tov on >40 cores

[Ian Hinder]

On 23 Feb 2017, at 15:47, Erik Schnetter <schnetter at cct.lsu.edu> wrote:

> Regarding OpenMP:
> 
> Cactus usually tries to optimize which treads run on which cores. If there are multiple independent processes running on a node, then Cactus must not do that, since this will slow down both applications a lot. In particular, must not set "CACTUS_SET_THREAD_BINDINGS=1" (not setting it or setting it to 0 is fine).
> 
> As Ian mentioned, you can either build without OpenMP support, or choose to use a single thread at run time, both will work.
> 
> At this time, it might be best to post your complete setup, i.e. all the options and scripts you are using to configure and build and submit and run, so that others can have a look and cross-check.
> 
> Of course, all of this is independent of any nans you encounter. Those are still due to a bug. I don't think it makes much sense at this point to debug this -- instead, you will want to run a larger simulations.

Hi Erik,

Are you suggesting that it is not worth us debugging the problem with Carpet which is giving wrong results if too many processes are used?  This could be triggered in other situations; we don't know what the origin is.

If you mean that Gwyneth would be better off focusing on larger simulations than trying to get this one to work on so many processes, then I agree.

-- 
Ian Hinder
http://members.aei.mpg.de/ianhin

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.einsteintoolkit.org/pipermail/users/attachments/20170307/87ef5b69/attachment.html