Re: [Gems-users] Accurate timing and performance simulation of x86machines


Date: Thu, 10 May 2007 09:06:18 +0200
From: "Fco. Javier Ridruejo" <franciscojavier.ridruejo@xxxxxx>
Subject: Re: [Gems-users] Accurate timing and performance simulation of x86machines
Thank you very much for your explanation.

Now we have a full system simulation environment to assess performance of
clusters inter-connected with custom networks. It is based on Simics plus an
interconnection network simulator of our own, but due to the lack of
accurate timing, results are not very accurate.
We thought of using GEMS to enhance the accuracy, but we also thought of
using GEMS to simulate NUMA machines inter-connected with our functional
simulator of interconnection networks (FSIN), maybe this last approach is
possible.
We will explore M5 to check if we can use it to get accurate results on
simulation of clusters with interconnection networks.

Greetings,

----------------------------------------
Francisco Javier Ridruejo Pérez
Red Académica i2BASK (UPV/EHU)
The Basque Country University
Parque Tecnológico de San Sebastián
Pº Mikeletegi, 69 - Torre Arbide Norte
20009 Donostia - San Sebastián
Tel.:  +34 943 018 705
Fax.:  +34 943 015 590
E-mail: franciscojavier.ridruejo@xxxxxx
----------------------------------------
-----Mensaje original-----
De: gems-users-bounces@xxxxxxxxxxx [mailto:gems-users-bounces@xxxxxxxxxxx]
En nombre de Mike Marty
Enviado el: Wednesday, May 09, 2007 9:39 PM
Para: Gems Users
Asunto: Re: [Gems-users] Accurate timing and performance simulation of
x86machines

GEMS is best suited towards modeling a shared-memory hierarchy.  It does not
handle I/O devices like Myrinet and Infiniband.

If you are not committed to Simics, I believe the Michigan M5 simulator
might be better suited to simulating a cluster with custom networks.

http://m5.eecs.umich.edu/wiki/index.php/Main_Page

--Mike


> Hello,
>
> I am Francisco Javier Ridruejo from The Basque Country University and 
> we are planing to use GEMS to enhance Simics to obtain an accurate 
> performance modeling of a cluster interconnected by a custom network 
> like for example Myrinet or Infiniband. Now we use plain Simics 
> simulating a Red Hat operating system over Pentium 4 machines.
> As far as I know Can we use Gems to accurate simulate the Simics 
> timing on
> x86 machines? Or GEMS only does accurate timing and performance for 
> Sparc machines?
> What is exactly the patch to convert the GEMS memory system module, 
> Ruby, to work with x86 target machines?
>
> Thank you very much in advance for your answer.
>
> Greetings,
>
> ----------------------------------------
> Francisco Javier Ridruejo Pérez
> Red Académica i2BASK (UPV/EHU)
> The Basque Country University
> Parque Tecnológico de San Sebastián
> Pº Mikeletegi, 69 - Torre Arbide Norte
> 20009 Donostia - San Sebastián
> Tel.:  +34 943 018 705
> Fax.:  +34 943 015 590
> E-mail: franciscojavier.ridruejo@xxxxxx
> ----------------------------------------
>
>
>
>
> _______________________________________________
> Gems-users mailing list
> Gems-users@xxxxxxxxxxx
> https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> Use Google to search the GEMS Users mailing list by adding
"site:https://lists.cs.wisc.edu/archive/gems-users/"; to your search.
>
>

_______________________________________________
Gems-users mailing list
Gems-users@xxxxxxxxxxx
https://lists.cs.wisc.edu/mailman/listinfo/gems-users
Use Google to search the GEMS Users mailing list by adding
"site:https://lists.cs.wisc.edu/archive/gems-users/"; to your search.



[← Prev in Thread] Current Thread [Next in Thread→]