[Cbe-oss-dev] PATCH [0/3] spusched enhancements

[Daniel Brokenshire]

Alex,

There is no formal test suite. However any application/workload can used 
to generate an overcommit situation. The three scenarios of interest are:

Two or more applications/workloads are run simultaneously, each of which 
does not overcommit the SPEs. 
One application/workloads that overcommits the SPEs. Many of our code 
samples allows one to specify the number of SPEs.
Combination of 1 and 2.

#1 is probably the most important scenario. 

Sincerely,

Dan B.
-------------------------------------------------------------------------------
Daniel A. Brokenshire
STSM / Emerging Products Development
brokensh at us.ibm.com               512-838-3373



Alex Chow/Austin/IBM at IBMUS 
Sent by: cbe-oss-dev-bounces+brokensh=us.ibm.com at ozlabs.org
10/24/2007 08:33 AM

To
Christoph Hellwig <hch at lst.de>
cc
Kanna Shimizu/Austin/IBM at IBMUS, Arnd Bergmann <arnd.bergmann at de.ibm.com>, 
cbe-oss-dev at ozlabs.org, Christoph Hellwig <hch at lst.de>, 
cbe-oss-dev-bounces+achow=us.ibm.com at ozlabs.org
Subject
Re: [Cbe-oss-dev] PATCH [0/3] spusched enhancements






Hi, is there a performance test suite to measure the overhead impact of 
"over committed" case?

Thanks..
=======
Regards,

Alex Chunghen Chow


cbe-oss-dev-bounces+achow=us.ibm.com at ozlabs.org wrote on 10/24/2007 
03:23:32 AM:

> Hi Luke,
> 
> I like this a lot.  I remember we had some problems with pagefault
> handling with this idea, so we might have to look into that again.
> 
> More detailed comments later.
> 
> On Tue, Oct 23, 2007 at 11:09:02PM -0300, Luke Browning wrote:
> > Hi Arnd and Christoph,
> > 
> > This patchset should significantly improve the performance of the 
system
> > when the SPUs are over committed. I have tested these changes on the
> > simulator with good results. I can now run several instances of the
> > matrix multiplication testcase at the same time in a few minutes. 
> > 
> > I haven't tested 'isolated' spu contexts yet. 
> > 
> > Do you have a testcase for 'isolated' contexts?
> 
> I think Jeremy's testsuite now has a testcase for it.
> 
> _______________________________________________
> cbe-oss-dev mailing list
> cbe-oss-dev at ozlabs.org
> https://ozlabs.org/mailman/listinfo/cbe-oss-dev
_______________________________________________
cbe-oss-dev mailing list
cbe-oss-dev at ozlabs.org
https://ozlabs.org/mailman/listinfo/cbe-oss-dev

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ozlabs.org/pipermail/cbe-oss-dev/attachments/20071024/f336ad3b/attachment.htm>