Hi folks. To try to speed up simulation, I was thinking of pulling some
unidirectional processing (ie things which leave gem5 and don't loop back
in) out of the critical path and farm them out to threads. I'm specifically
thinking of the code that creates framebuffer images, and potentially code
that feeds things to VNC. This is particularly important when the
simulation is otherwise sped up with, for instance, KVM or Fast Models.

Since some output will be coming from threads which may not be finished
when gem5 is ready to exit (or crashes, or ...), I want to be able to
ensure that gem5 will wait for them before shutting down. I wouldn't want
to lose information that might be critical for debugging a crash because it
was being written out asynchronously.

Is there any way to hook into something like that? I'd just want to join
any threads which are not yet finished running. Or is this something the
std::thread implementation will do for us automatically? That would be
nice, but I doubt it!

Gabe
_______________________________________________
gem5-dev mailing list -- [email protected]
To unsubscribe send an email to [email protected]
%(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s

Reply via email to