Folding Forum

Ok the Server is up and running !! Im interested in bigadv, i think this proliant should work great for that. Here are the spec's:
Hp Poliant DL858 G6
CPU : 4 x 6core AMD Opterton 8431
CPUID : Family 16 Model 8 Stepping 0
RAM : 32gb (not sure speed)
Os : Win Server 2012 Standard

Right now the client is telling me that PPD is 29848 and slowly rising. I feel like this is still very low, my personal pc will run 40k+PPD... What do you guys think im doing wrong?

Another Question that comes to mind; Can i link all 4 servers to the same project to complete them faster?

mark5365 wrote:Another Question that comes to mind; Can i link all 4 servers to the same project to complete them faster?

No, the F@H client and folding cores do not support clusters. Nor do the assignment servers support just assigning WU's from a single Project to a group of systems. Attempts to do so on your own would be a violation of the license and the Best Practices quide.

mark5365 wrote:Right now the client is telling me that PPD is 29848 and slowly rising. I feel like this is still very low, my personal pc will run 40k+PPD... What do you guys think im doing wrong?

Can you go over to the logs and post a large chunk of it here, one that includes both the project number and several percent of its progress? The PPD estimate that the client provides isn't always accurate. Actually, you could double-check it yourself by going here and inputting the project number and average frame time.

Also check Task Manager to see if there are any other processes consistently using CPU time. Multi-core folding only progresses as fast as the slowest thread; another task continually taking up even a few percent of one core can drastically increase frame times.

You'll have to setup and use the -bigadv flag to get bigadv work units if you haven't already.
As stated you either have something using CPU cycles other than folding or a standard
smp work unit working some of which can have some low PPD. Best to post the Project number
its running and some of the time data from that project. Then they can tell you more.

mark5365 wrote:Ok the Server is up and running !! Im interested in bigadv, i think this proliant should work great for that. Here are the spec's:
Hp Poliant DL858 G6
CPU : 4 x 6core AMD Opterton 8431
CPUID : Family 16 Model 8 Stepping 0
RAM : 32gb (not sure speed)
Os : Win Server 2012 Standard

Right now the client is telling me that PPD is 29848 and slowly rising. I feel like this is still very low, my personal pc will run 40k+PPD... What do you guys think im doing wrong?

There is no Windows bigadv WU's, bigadv is LInux only so you will not be able to run Bigadv unless you chang to a Linux OS. As far as your current PPD goes are your running -smp 24 have you checked to see if the cpu's are 100% loaded.

The proliant client is at 35463 PPD now and all cores are maxed out @ 100%. i still feel like running a linux VM box would make this faster. I will look around today and see what i can find.

mark5365 wrote:...i still feel like running a linux VM box would make this faster...

Yes.

Linux VM > Windows, native Linux > VM.

Amaruk wrote:

mark5365 wrote:...i still feel like running a linux VM box would make this faster...

Yes.

Linux VM > Windows, native Linux > VM.

Getting a linux VM to efficiently use a lot of cores is not trivial though. Unless you need bare metal access in windows, running native linux with a windows VM is a better option from a folding perspective.

The reason im running windows server 2012 is so that i can learn and get used to all the changes. I own decent sized computer, tablet, smartphone repair shop so i need to learn as much as i can without stepping in a classroom. Bad part is most all medium sized businesses are only running windows servers or have only heard of windows so learning linux outside of a VM is not good for me.

Another Update
So i picked up 2 more ProLiant DL585 G5 (only 4x 4core Opertrons) and shut down the dell PowerEdges. So now i have 24/16/16/8/4 core machines in this farm Folding@home

I will start working on The VMware tonight and hope to have everything working by the end of the weekend.

Dose anyone have a save for linux with F@H already installed that i can use for the VMware? Long time ago i have one that was super simple to setup but i dont remember what the name of it was. Any help here would be greatly appreciated to speed this process up. Time is hard to come by and i have already put about 50hrs this week into get the servers, installing them, and loading them. On top of working 50hrs....

Thank again,
Mark

I am not sure if that VM could be distributed since the EULA states that the Client can only be downloaded from Stanford's site (and it's partners). Thus, having the F@H Client installed in the VM being distributed would be a violation of the EULA.

what am i doing wrong here? 80% for a total of 44k DDP... still seems very low for 24 cores...

Here is fahmon... this is not going to make the deadline... im really feeling lost...
here are the other 2 16 core servers. they have the same specs not sure why they are not doing the same PPD...

In the first screenshot you posted, it's not starting up three copies of the client, is it? The text after "Manual Startup" shows three Start commands, but I can't tell if it's running them all or if it's just echoing different options. If it is that, figure out how to only run the bigadv instance and not the other two. If it's not that, then check to see if there are any other tasks using up a lot of CPU time.

Is the machine in the first screenshot the 24-core server? If so you might try changing the "-smp" flag to "-smp 24". I'm not sure if the v6 client by default will attempt to handle that many threads.

As to the 16-core servers, I would look at the log files to see what the actual TPF's are. v7's PPD estimates are known to be off sometimes. I would have said "different projects have different PPD" but it looks like they're both running 7808's. It's not unknown for different WUs from the same project to run with different TPFs on the same hardware, though, and I can't make out how different they are from that screenshot.