Re: [Gems-users] Processor binding for LogTM

[Shantanu Gupta <shangupt@xxxxxxxxx>]

Thanks Jayaram for the response. Actually I want to be clear about this 
fact, could you elaborate a little more ? I want to know if binding is:
1) recommended, OR
2) absolutely required for the correct execution

Because even for the SPLASH2 benchmarks I am using, I did not add any 
processor thread binding.
-Shantanu

On Thu, 26 Jul 2007, Jayaram Bobba wrote:

> The distributed version of LogTM does not handle thread switching and
> paging.
> Hence thread binding is recommended.
>
> Jayaram
>
> Shantanu Gupta wrote:
> > For correct execution of a multithreaded benchmarks using logTM protocol,
> > is processor binding (pset_bind) necessary ?
> >
> > If yes, then why does the deque benchmark not have it ?
> >
> > If no, then how is it guaranteed that a thread will be not be shuffled out
> > of a processor while it is running under a transaction ? What is
> > responsible for ensuring this :
> >    - logtm coherence protocol
> >    - logtm transaction manager
> >    - simics interface, etc..
> >
> >
> > -Shantanu
> > _______________________________________________
> > 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.