[PTLsim-devel] PTLSim on AMD-64

Wed Apr 16 04:59:58 EDT 2008

The problem was instability, with multicore FullSystem simulations. With
some revisions even vanilla version of PTLSim didn't work in multicore. With
some it did, but for all of them, when we add our changes, the simulator
blocks (doing 0 commits per second), or in some cases deadlocks, after some
time doing simulations. It was usually between 1 and 30 minutes before the
problem appears.
The problem could be caused by numerous "illegal instruction" calls - that's
what we've been thinking recently. We did a lot of these calls, like 500.000
.
Anyway, this problem will probably last while PTLSim is bounded to Xen. We
hope that will change soon.

BTW, the multicore simulation is *not yet* cycle accurate! And if you're
doing single core full-system simulations you're probably OK, with both
stability and accuracy.

regards!

On Wed, Apr 16, 2008 at 3:44 AM, Tommy <tommy24 at live.com> wrote:

>  Hi Sasa,
> Can I ask you what problem you have on Xeon?
> Because I already installed ptlsim on Xeon machine, I would like to know
> what problems you have then I can check to see if I have the same problem.
> So far, I didn't see any problem although I didn't run any benchmarks
> yet. I appreciate that if you can tell me some details about the problem you
> have.
> Thank you.
>
> - Tommy
>
>  *From:* Saša Tomić <sasa.tomic at bsc.es>
> *Sent:* Friday, April 11, 2008 11:49 AM
> *To:* George Riley <riley at ece.gatech.edu>
> *Cc:* ptlsim-devel at ptlsim.org
> *Subject:* Re: [PTLsim-devel] PTLSim on AMD-64
>
> Sorry, I'm not in the PTLSim development team :)
> I was telling that from my personal experience, since we also had a lot of
> problems running PTLSim on our Intel Xeon machines, and we're also planning
> to switch to AMD64 in near future.
>
> I hope that somebody from the PTLSim dev-team can give us all hints about
> this?
>
>
> On Fri, Apr 11, 2008 at 5:15 PM, George Riley <riley at ece.gatech.edu>
> wrote:
>
> >  Sasa,
> > THanks for the info below.  We have been struggling without success to
> > get this working on an Intel Dual-Core Duo.
> > Can you tell me which motherboard you used for the AMD machine?  There
> > are several available, each with different
> > chipsets, and I want to make sure we get one that is known to work.
> >
> > Thanks!
> > George
> >
> >
> > For stability issue (Xen interface is _very_ fragile), I would *
> > personally* strongly suggest you to replicate their work environment.
> > Meaning, AMD64 machine with openSuse 10.2. And you should dedicate this
> > machine to PTLSim, because you'll need to do (relatively) frequent reboots.
> > good luck!
> >
> >    --------------------------------------------------
> > George Riley
> > Associate Professor
> > Georgia Tech
> > Electrical and Computer Engineering
> > riley at ece.gatech.edu
> > 
> > Klaus Advanced Computing Building
> > Room 3360
> > 266 Ferst Drive
> > Atlanta, Georgia 30332-0765
> >
> > ECE4110 Web page:
> > http://users.ece.gatech.edu/~riley/ece4110/>
> >
> > ECE4112 Web page:
> > http://users.ece.gatech.edu/~riley/ece4112/>
> >
> >
> >

-- 
Saša Tomić
BSC - Barcelona Supercomputing Center
c\ Gran Capita 2-4, Nexus I/204, 08034 Barcelona, España
Tel.: , web: http://www.bsc.es
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://ptlsim.org/pipermail/ptlsim-devel/attachments/20080416/a81f5ed9/attachment.htm 