In fact, there are some problems with 2 benchmarks (cholesky and
lunoncontig) even with the updated script. These benchmarks require us to
pass all ddefault arguments, otherwise they dont end correctly (you can pass
the extra -t arg to check results). I'll post a fixed config file that I
use.
- Sujay
----- Original Message -----
From: "Steve Reinhardt" <[EMAIL PROTECTED]>
To: "M5 users mailing list" <m5-users@m5sim.org>
Sent: Sunday, December 02, 2007 6:56 PM
Subject: Re: Fwd: [m5-users] CMP simulation
Note that there were some issues with the splash script in passing
args to splash benchmarks, and Ali posted an updated version to the
list recently.
Steve
On Dec 2, 2007 3:19 PM, Sujay Sunil Phadke <[EMAIL PROTECTED]> wrote:
you can execute any benchmark in the splash2 suite by the "-p" option
which specifies the number of processors.
~<....> -p 4 - b FFT
does this do what you want?
- Sujay
On Sun, 2 Dec 2007, [EMAIL PROTECTED] wrote:
> Yes,, you are right, my stats shows that no instruction is executed in
> the
> second cpu. I was using FFM, is there any benchmark in splash2 that
> would
> utilize both processors?
>
> -panos
>
> Quoting Steve Reinhardt <[EMAIL PROTECTED]>:
>
>> Oops, meant to send this to the list and not just to Ali.
>>
>>
>> ---------- Forwarded message ----------
>> From: Steve Reinhardt <[EMAIL PROTECTED]>
>> Date: Dec 2, 2007 12:50 PM
>> Subject: Re: [m5-users] panic on partially satisfying the functional
>> request
>> To: Ali Saidi <[EMAIL PROTECTED]>
>>
>>
>> No, they're not. The functional accesses Panos is seeing are probably
>> from syscall emulation.
>>
>> If there aren't any traces from the second CPU's cache, that probably
>> means the application is never getting to the point of creating a
>> thread on the second CPU.
>>
>> Steve
>>
>>
>> On Dec 2, 2007 11:36 AM, Ali Saidi <[EMAIL PROTECTED]> wrote:
>>> Actually, now that I think about this I'm not sure that it is? Are
>>> express snoops functional accesses?
>>>
>>> Ali
>>>
>>>
>>> On Dec 2, 2007, at 2:33 PM, Ali Saidi wrote:
>>>
>>> > They're "express snoops," take a look at:
>>> http://www.m5sim.org/wiki/index.php/Coherence_Protocol
>>> >
>>> >
>>> > Ali
>>> >
>>> > On Dec 2, 2007, at 11:55 AM, [EMAIL PROTECTED] wrote:
>>> >
>>> >> Hi I am reposting my question. I am running splash2 with 2 CPUs.
>>> >> Although I use the trace cache flag option, on the output I see no
>>> >> traces of my second's CPU cache.
>>> >>
>>> >> Moreover, since I have configured my system for atomic requests,
>>> >> who is calling the functional requests on the bus?
>>> >>
>>> >> Thanks
>>> >>
>>> >> Panos
>>> >> _______________________________________________
>>> >> m5-users mailing list
>>> >> m5-users@m5sim.org
>>> >> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
>>> >>
>>> >
>>> > _______________________________________________
>>> > m5-users mailing list
>>> > m5-users@m5sim.org
>>> > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
>>> >
>>>
>>>
>> _______________________________________________
>> m5-users mailing list
>> m5-users@m5sim.org
>> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
>>
>>
>>
>
>
> _______________________________________________
> m5-users mailing list
> m5-users@m5sim.org
> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
>
_______________________________________________
m5-users mailing list
m5-users@m5sim.org
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
_______________________________________________
m5-users mailing list
m5-users@m5sim.org
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
_______________________________________________
m5-users mailing list
m5-users@m5sim.org
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
