And we are glad to see your return. Crunch on with all the capacity you can spare.

Cheers

Hey MrKermit,

Is there a link where we can see what's under the hood?

Not really a link... the current crunch is on a pile of 6 year old Dual Socket Nehalem 2.93Ghz nodes with 8 cores and 24GB RAM. I may get as many as 384 onto WCG for a bit but it's probably about 1/2 that now. All with DDR Infiniband [20Gb] networking which WCG doesn't use but enables Supercomputing apps to run across them all with ~1.5 microsecond latency for truly parallel performance Using 'MPI' programming techniques/apps.

Yes, I pay my own power bills, no University mis-appropriations of compute labs!

We will have more soon. Luckily with the Config we have we run "Unlimited" CEP2 jobs on each node. That seems to hurt us at the "Total Results Returned" but our points and hours are in the top 5 which is fun.

I am curious what Mamajuanauk is running to hit nearly 2TF. I'm impressed by efficieny more than time spent, and 196Pts/h is awesome. We have hyperthreading enabled which isn't helping efficiency but does help put more of our RAM to work and gets a bit more throughput.

Ultimate we run big research computers, and ethically they need to do research for their entire lives to justify their existence. So if they aren't under contract we are donating their time to someone, and BOINC WCG makes that pretty easy if no one needs our interconnect for traditional High Performance Compute research work at that instant.

Congrats MrKermit for breaking 10Tflops today!
Very impressive!