* [OpenMP] [libgomp] Limit the number of threads spawned around
@ 2019-10-23 16:02 Quaquaraquà
0 siblings, 0 replies; only message in thread
From: Quaquaraquà @ 2019-10-23 16:02 UTC (permalink / raw)
To: gcc-help
Hi there,
I'm an external user of a target library written in C++/OpenMP. The library has its own API and it claims to be thread-safe. Internally, its operations are parallelised with the usual #pragma omp parallel for. I'm deploying it on a server with 40 physical threads and invoking the library from 40 (active) logical threads. This creates 40*40=3600 logical threads in the machine. My question is whether there is a way, as external user of this target library, to limit this insanity so that the OpenMP threads use the resources available in a more composed manner (e.g. worker pool) or should I limit myself to interact with this library from a single logical thread?
Thanks in advance,
Yours,
quaquaraqua
--
Quaquaraquà <nuquaquaraqua@gmail.com>
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2019-10-23 16:02 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-10-23 16:02 [OpenMP] [libgomp] Limit the number of threads spawned around Quaquaraquà
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).