folding-at-home team mailing list archive

[Steve Woodruff <swoody@xxxxxxxxxx>]

Dan, (sorry, mailing list fail) ;)

Thanks for the great info on the new A3 core! I just wanted to mention
a couple quick points about this new work unit as well.

First, as you've said, some people are getting higher PPD with this
new client, however for some, it's actually producing less than the A2
work units. It seems Stanford is in favor of processors with high
cache and not just aimed at pure GHz. This has been pretty noticeable
with some of their latest work units, and is apparent in the new A3
ones as well.

Stanford also seems to be aiming for higher end processors as of late.
This became apparent with the new -bigadv work units, and again will
be continuing with the new A3 work units. This is really great since
Stanford can now assign 'science critical' units to those who have
higher end systems, and know that they will be able to get results
back very quickly.

The bonus point system for the -bigadv work units will also be used
for the A3 units. This means that you get more points for how quickly
you complete and turn in your work unit. I think this is a great idea
for those who have much more powerful systems, and are constantly
completing units rapidly. This seems to be something which they're
considering for all future work units, and I hope they stay with this
trend.



--
Steve Woodruff
swoody@xxxxxxxxxx



On Thu, Jan 28, 2010 at 11:21 AM, Daniel Chandler <djchandler@xxxxxxxxxx> wrote:
> Some of you may have noticed your PPD changing or machines running
> slower, especially those running smp WUs. There are several reasons.
>
> If you are using an older client, 6.24 or older, you may be getting WUs
> that require the older CoreA1 now. CoreA1 has markedly higher times per
> frame and under utilizes 4 core CPUs as compared to CoreA2. CoreA3 will
> only run under the new 6.29 client. CPUs equivalent or faster than a
> stock Q6600 could higher PPD when running CoreA3, and faster dual cores
> with level 3 cache could benefit as well.
> http://foldingforum.org/viewtopic.php?f=44&t=13099
> http://foldingforum.org/viewtopic.php?f=44&t=13045
>
> Kasson @ Stanford can explain this far better than I. So check out these
> links.
> http://foldingforum.org/viewtopic.php?f=24&t=12997
> http://foldingforum.org/viewtopic.php?f=24&t=13038
> http://foldingforum.org/viewtopic.php?f=24&t=13039
>
> Here's a couple of other sites for news.
> http://folding.typepad.com/news/
> http://en.fah-addict.net/news/news.php
>
> Dan
>
>
>
>
>
>
> _______________________________________________
> Mailing list: https://launchpad.net/~folding-at-home
> Post to     : folding-at-home@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~folding-at-home
> More help   : https://help.launchpad.net/ListHelp
>