PDA

View Full Version : Where is the fabled A2 core?


orion456
09-10-08, 06:18 PM
What happened to the 100% efficient A2 core? Talked about but never seen?

SMP folding looks pretty lame these days especially when 2 cores fold almost as fast as 3 now. :bang head

Macaholic
09-10-08, 06:34 PM
On Linux and OS X. You can see it here (http://foldingforum.org/viewtopic.php?f=44&t=4155). ;) Not too sure about Winders, however. That is quite up in the air.

dicecca112
09-11-08, 06:08 AM
yup and you need the advmethods flag set. Those WU are great, especially on a quad.

harlam357
09-12-08, 08:54 AM
Yep... I'm getting ~6k PPD from a Q6600 @ 3.4GHz using two Ubuntu VMs, two cores per- no GPUs. :)

orion456
09-14-08, 01:30 PM
Yep... I'm getting ~6k PPD from a Q6600 @ 3.4GHz using two Ubuntu VMs, two cores per- no GPUs. :)

Now why did I think the only reason you needed two VMs was the inefficiency of the old cores. Shouldn't you get more points now with all 4 cores operating on one WU? :shrug:

David_N
09-14-08, 01:44 PM
Linux folds better, and VM's only use two cores each for the free version. Also, I believe the SMP process is optimal with two cores.

Sorin
09-14-08, 02:35 PM
Also, I believe the SMP process is optimal with two cores.

That's just funny, considering it's hard coded for a quad (four threads). :p

harlam357
09-14-08, 03:01 PM
Now why did I think the only reason you needed two VMs was the inefficiency of the old cores. Shouldn't you get more points now with all 4 cores operating on one WU? :shrug:

I have no idea... all my quads are setup with two VMs and I'm going to keep it that way for the foreseeable future. I'm a Windows guy, but I use Linux to Fold since, as David_N points out, it's a bit faster and more stable than WinSMP. In fact, I only have one WinSMP client running on my farm anymore... on my 5000+ BE, which is my HTPC box.

That's just funny, considering it's hard coded for a quad (four threads).

Just want to make the distinction that the SMP client is multi-process, i.e. each Fah_Core executable, as well as multi-threaded, within each core process. ;)