SMP2 core update

Soldato
Joined
17 Jan 2007
Posts
8,944
Location
Manchester
We are excited to announce progress in our testing of new SMP2 cores. These cores use threads-based parallelization instead of MPI and, we hope, should be much more robust than our first-generation SMP cores.

The SMP2 cores are currently in testing; watch for future announcements regarding their release. The first SMP2 core to be released will be a Gromacs-based core with core number A3. In conjunction with this release, we are implementing some updates to our points benchmarking system. In particular, we will use early-completion bonuses, which we have been testing with the bigadv work unit program, more extensively for SMP2 cores.

http://folding.typepad.com/news/2009/12/smp2-core-update.html

Early completion bonus sounds nice.
 
I like the sound of more robust and no MPI crap. Might be able to run it natively in Windows without the ball ache.

Early completion bonus sounds awesome! Teach those damn slackers to buy better hardware. :p
 
Sounds promising, does it give an expected date of release in the article? I can't open it in work.
 
Having read a few threads on their forums it seems both the SMP2 and GPU3 clients are 'weeks' away rather than months, that is a specific as they are being.

Its typical this happens the day after i finally get a stable linux VM - never using notfred again! All my effort may now be due for obsoletion:rolleyes:
 
notfred seems ok for me in my tests the last few days, but a windows smp client that has equivalent throughput the to the Linux one would be much less pain :)
 
Although I took the time to get VMware set up I barely use it now as the computers off and on so much. Windows SMP just isn't worth running for me at the moment so will be looking forward to this if it's any good.
 
It would solve a problem for me if the new Windows SMP client is at least stable and works. I don't mind a small PPD drop compared to Linux if it means I avoid the hassle of VMs running on my main PC all the time!
 
Heh, whats up with you guys, I think you are pushing your overclocks too hard.

Windows SMP as it stands has been stable as anything for me but only gets a max of 2KPPD on a Q9650 whereas the VMware notfred client gets nearer 5KPPD

But both run smooth as. Windows client is less of a PITA because it runs as a service.
 
I'm with biffa on this one - but them I over clock at or below stock volts.
At stock or 10% OC (my work farm when I ran it) was rock solid stable on windows smp as a service.
Even had them all running a web based Fahmon alturnative, again as a service to check progress without leaving my workstation.
Was a sweet farm - then got the electric bill and that was the end of that.

Just ran the C2D's work hours after that - then gpu folding arrived and I was back in the top 20 producers again :)
 
Last edited:
My overclock is fine :p (IBT, Prime and memtest stable - i can't force it to fall over!).

The WinSMP client is fine for me too, it just nets ~2k ppd.
I moved to a notfred client ~5kppd, and have been having random lockups - this could be minutes or days.
I've just moved over to a proper ubuntu VM install and the random lockups have gone, which is why i wont be using notfred again. The ubuntu client also seems to be running slightly faster :)

Capture.jpg
 
Last edited:
I just could never get the Windows SMP client to install! And wow, I never realised it was 2-3K down on PPD!

Screw Windows, not losing 3K PPD for convenience! :p

On a related issue, as you will know from my SBS thread I maaaay build a couple of dedicated CPU folders. I have been out the hardware loop for a while, so could one of you guys take a look at my proposed spec in that thread and inform me of any better solutions. Better keep it in that thread too, don't want to derail this one :)
 
Windows SMP is not only down on PPD, but for me it always lost work if the PC had to be rebooted. It also absolutely knackered system performance (had to take it off a server because other more important background tasks were taking half an hour instead of a minute or two).
 
Back
Top Bottom