[gmx-users] 8MB CPU cache vs 4MB CPU cache: large difference in performance?

[David van der Spoel]

Yang Ye wrote:
> Cache size matters because it ensures a steady data/instruction flow to 
> feed the "hungry" CPU, especially with Intel Coreo architecture. With 
> database-type application, larger cache benefits much. In case of 
> GROMACS, unless you deal with very large system, I doubt its 
> effectiveness. It could be better to run some real-world benchmarks. Or 
> look for benchmark that contains software operating in a similar way of 
> GROMACS, e.g. scientific benchmark, multimedia benchmark (decoding/3D 
> rendering, with SSE boost).

but note that you now get four cores that share the cache, so 1 Mb/core 
or 2Mb/core. That does matter. For an honest comparison you should 
compare (price with 8Mb/price wih 4Mb) / (Performance 8 Mb / Performace 
4 Mb), where performance is in ns/day. I would think that 60$ extra per 
chip is worhtwhile.

>  
> Regards,
> Yang Ye
> 
> 
> ----- Original Message ----
> From: Frankie Montenegro <frankie.montenegro at gmail.com>
> To: gmx-users at gromacs.org
> Sent: Sunday, September 9, 2007 7:08:54 AM
> Subject: [gmx-users] 8MB CPU cache vs 4MB CPU cache: large difference in 
> performance?
> 
> Hi,
> 
> I am about to purchase several units with Intel's quadcore q6600
> processors. Since about 80% of our work is done with Gromacs, I
> figured this list is a place to ask performance questions.
> 
> I was wondering if there would be any significant benefit in having a
> q6600 processor with 8MBof L2 cache over the identical processor
> with 4MB of L2 cache. The price difference is about 60$,which doesn't
> seem much,  but could pile up if we end up buying more then a few units.
> 
> Thanks for your help.
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before posting!
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before posting!
> Please don't post (un)subscribe requests to the list. Use the 
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php


-- 
David van der Spoel, Ph.D.
Molec. Biophys. group, Dept. of Cell & Molec. Biol., Uppsala University.
Box 596, 75124 Uppsala, Sweden. Phone:	+46184714205. Fax: +4618511755.
spoel at xray.bmc.uu.se	spoel at gromacs.org   http://folding.bmc.uu.se