[mvapich-discuss] (no subject)

[Sayantan Sur]

Hi,

* On Mar,7 Troy Telford<ttelford at linuxnetworx.com> wrote :
> On Thu, 23 Mar 2006 02:13:27 -0700, Roland Fehrenbacher  
> <Roland.Fehrenbacher at transtec.de> wrote:
> 
> >    >> When running HPL, hpl.dat can contain multiple problem sizes.
> >    >> xhpl * reads the config file, and runs one problem size (until
> >    >> it completes the problem) * After the previous problem is
> >    >> finished, hpl will then start execution on the next problem
> >    >> size.
> >
> >    Sayantan> Thanks for the explanation. I understand what you are
> >    Sayantan> saying.
> >
> >I'm facing the same problem here with xhpl and mvapich 0.9.7. The
> >strange thing is that this problem didn't happen with versions at
> >least up to 0.9.5 (haven't tested 0.9.6). Is the mechanism mentioned
> >below a performance optimization that entered the code after 0.9.5? I
> >just checked, and I had set -DLAZY_MEM_UNREGISTER in my 0.9.5 version
> >already.
> 
> I've actually been able to replicate it with 0.9.5.  (And even 0.9.4)

Troy, thanks for verifying this behavior with 0.9.5 and other previous
releases. As mentioned in the previous email, this is because of the way
malloc needs to be configured in order to cache registration entries.

Roland, this optimization has been in MVAPICH for quite a long time
(ever since the earliest releases). In our recent release 0.9.7, the
registration caching algorithm was optimized and a user configurable
limit on the max. number of registered pages was provided (since 0.9.6).
However, the basic mechanism (ie. configuring malloc) remains the same.
Would you describe your experimental setup when you saw a different
behavior with 0.9.5?

Thanks,
Sayantan.

> -- 
> Troy Telford

-- 
http://www.cse.ohio-state.edu/~surs