Points System for SMP2 WU

[phil3_66666]

Originally Posted by geokilla

Are you including the bonus in your calculations? I'm going to make the change soon and will probably have to install HFM.NET since FahMon doesn't calculate the bonus.

with bonus, use hfm.net...


i can confirm for dual core, even at 3.85ghz it's too slow

[phil3_66666]

Originally Posted by Dead Things

Last night I spent a lot of time testing the new WUs out on dual core and quad core hardware from both AMD and Intel, comparing results of the new SMP client for Windows and the A3 core versus the previous SMP client for Linux under VMware and the A2 core. And I feel like I've seen enough to say this:

- Both AMD and Intel quads, but especially Intel quads are faster with the new SMP client and A3 cores.

- i7 processors with HT are faster with the new SMP client and A3 cores when compared to regular SMP projects, but bigadv projects (which still use the A2 core) continue to outperform the new client and core.

- Both AMD and Intel dual cores are slower with the new SMP client and A3 cores. They are not fast enough to earn enough of a bonus. As a result (for the timebeing at least), dual core users should stick with their current configurations.

on my side i also tried the 3core athlon 425 at 3.6, with an useless core in vmware it give me 2100ppd, with smp2 it's now running on 3 core at 3100ppd(with bonus)

[Zero82z]

Originally Posted by Dead Things

Yes - Kasson has instructed Bigadv folders to update their clients (hence the new FredNotFred and dev version of LinuxRouter's appliance). That can only mean that A3 is coming to Bigadv at some point.

-bigadv will be coming to SMP2, but that wasn't the reason for the client update. There was a critical issue with the old client version and the A2 core -bigadv units which was fixed in 6.29. That's why the update was made mandatory for -bigadv folders.

Originally Posted by Dead Things

Here's the thing though. As I understand it, A3 behaves similarly on both Windows and Linux. So, if Bigadv will soon be migrated to A3, then one might speculate that Bigadv will also be available on the new Windows SMP client. And since a native installation will always outperform a virtualized one, we could be in for some big time numbers from Bigadv.

Nothing official has been announced yet, though, and the concept of Bigadv running on the Windows SMP client is purely speculation.

I'd love it to be true though, 'cuz then I could actually use the four PCIe slots on Leviathan! (once it comes back from RMA. sadface.)

I see no reason why A3 -bigadv units would not work with the Windows SMP client, since the SMP2 core is the same regardless of the OS.

[Charloz24]

I finally move my Q6000 @ 3hgz from 2 VWmare to just one SMP2 win installation! It was indeed a good move, was 4000 ppd before and now 6000+ ppd easy (and just one instance running)

Question now.... in the next days I plan on adding a 9800 GX2 and of course it will fold! What I do with the smp2 installation? Can I run the SMP2 client on 3 cores, for being able to have one core free to "juice" the GX2 folding or what? And will I need 2 GPU folding instances?

[lowfat]

Originally Posted by Charloz24

I finally move my Q6000 @ 3hgz from 2 VWmare to just one SMP2 win installation! It was indeed a good move, was 4000 ppd before and now 6000+ ppd easy (and just one instance running)

Question now.... in the next days I plan on adding a 9800 GX2 and of course it will fold! What I do with the smp2 installation? Can I run the SMP2 client on 3 cores, for being able to have one core free to "juice" the GX2 folding or what? And will I need 2 GPU folding instances?

Nvidia GPU's require almost zero use from the CPU. You can still run the SMP w/ 4 cores w/ very little issues.

[Charloz24]

well thanks thats very good news!

[Zero82z]

Originally Posted by Charloz24

Question now.... in the next days I plan on adding a 9800 GX2 and of course it will fold! What I do with the smp2 installation? Can I run the SMP2 client on 3 cores, for being able to have one core free to "juice" the GX2 folding or what? And will I need 2 GPU folding instances?

You might need to use an affinity changer program like Bill2's Process Manager to lower the priority of the SMP client to make sure it doesn't stifle the GPU clients, but that's it. You can still run SMP on all four cores.