Hi Jimmy,
I'm not an expert on cluster management, but I'll try to answer your questions anyway.
On 6/3/19 5:48 PM, Jimmy Mulder wrote:
*So, the questions:*
- Does NEST prefer CPU or GPU? Is it RAM intensive? I couldn't find
this on your website, or anywhere really. Are there ballpark numbers for the memory required for 10^9 connections?
This depends on the type of connection you are using, but you can assume each connection will consume somewhere between 8 and 32 bytes.
- Am I abusing the "local_num_threads" mechanic by applying it to a
cluster like this? Why is 1 thread sometimes faster than 10 threads?
first part: no, this is the intended use for a shared memory architecture. second part: context switches can kill performance if you are many more threads than you have cores available
- Would switching to MPI help?
I don't think so, but you should try!
- If so: the documentation says I need to run
$NEST_SOURCE_DIR/configure --with-mpi, but there is no "configure" folder or command to be found anywhere, and this line fails. What and where is this "configure"?
you are most likely using a recent NEST version for which you use cmake to configure before building. just pass -Dwith-mpi=ON when calling cmake, and check that in the build summary it reports that MPI is actually linked.
Hope this helps, Jakob