Page 1 of 1

bigadv move to 16+ cores as previously announced

PostPosted: Mon Jul 30, 2012 5:03 pm
by kasson
We announced at the beginning of the year that the bigadv program would be moving to clients with 16+ cores. At this point, we are only serving 16+ BA work units. We do continue to evaluate the program and may make changes (in either direction) in the future as both the scientific work we are doing and the capabilities of donor machines continue to change.

Thanks for folding!

Re: bigadv move to 16+ cores as previously announced

PostPosted: Mon Jul 30, 2012 8:20 pm
by Xavier Zepherious
so nothing for 3930k even tho I can do 6901,6903,6904 easily and should easily make 6905,6906

or are we still gonna have some wu's for such systems

Re: bigadv move to 16+ cores as previously announced

PostPosted: Mon Jul 30, 2012 8:25 pm
by Nathan_P
Xavier Zepherious wrote:so nothing for 3930k even tho I can do 6901,6903,6904 easily and should easily make 6905,6906

or are we still gonna have some wu's for such systems


I reckon I can manage 8102 on dual L5640's but not 8101 and probably not 6905/06.
Oh well off to the scrap heap it goes :cry:

Re: bigadv move to 16+ cores as previously announced

PostPosted: Mon Jul 30, 2012 8:29 pm
by 7im
Xavier Zepherious wrote:so nothing for 3930k even tho I can do 6901,6903,6904 easily and should easily make 6905,6906

or are we still gonna have some wu's for such systems


No, not nothing. As announced many months ago, less than 16 core performance would best help the project running regular SMP work units. ;)

Re: bigadv move to 16+ cores as previously announced

PostPosted: Mon Jul 30, 2012 10:30 pm
by Patriot
http://folding.typepad.com/news/2011/11 ... -2012.html

Announced Nov 14th 2011.
First BA16 wu much much later...

Re: bigadv move to 16+ cores as previously announced

PostPosted: Tue Jul 31, 2012 1:42 am
by Leonardo
Announced Nov 14th 2011. First BA16 wu much much later...
Well, we can't complain that we didn't get advance warning. :P

Re: bigadv move to 16+ cores as previously announced

PostPosted: Tue Jul 31, 2012 8:53 am
by rhavern
Leonardo wrote:
Announced Nov 14th 2011. First BA16 wu much much later...
Well, we can't complain that we didn't get advance warning. :P


But some probably will :-)

Re: bigadv move to 16+ cores as previously announced

PostPosted: Thu Aug 02, 2012 8:46 pm
by heikosch
I´d like to express that I´m a big fan of this change. In the statistics there are now some teams (just trying to overtake my team) with a large decrease of PPD/day, some of their folders are even reduced to 0 PPD/day - probably stopped folding. Strike! :-)

But I´ve still no idea how science could benefit from that.

Heiko

Re: bigadv move to 16+ cores as previously announced

PostPosted: Thu Aug 02, 2012 9:09 pm
by bruce
heikosch wrote:But I´ve still no idea how science could benefit from that.


Simple answer: FAH needs to maintain a balance for all projects. Managing the PPD rules is an attempt to do that, but if 50% of the CPUs are working on 15% of the projects and the other 85% of the projects need more CPUs, they can ask nicely, they can add 10% to Core_a4, and they can reduce the incentives for Core_a5. Good science is not achieved unless there's a reasonable balance between ALL research projects.

Some will resist that change and some will get mad and leave, but if it makes for a better balance, then it's good for FAH's research efforts, depending, of course on how radically it was out of balance. (I have no numerical data, just a qualitative statement.)

Re: bigadv move to 16+ cores as previously announced

PostPosted: Fri Aug 03, 2012 12:53 am
by PinHead
As with managing anything, you align the resources with the need. This is research so the need changes, or maybe someone just needs to do a sanity check on direction, or maybe we're not making the progress that we anticipated what can we try different. The list is endless, but it is research and if 50,000 ways that it doesn't work are discovered; then we are closer to the way it does work!