• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Linux SMP Help

L

lelliott731

[H]ard DCOTM April 2017
Joined
Mar 15, 2002
Messages
941
Hey guys, I need help!!!

Here is my setup:
Supermicro X7DCL-i (Intel 5100 Chipset) Motherboard
(2) Xeon E5430 (Quad Core 2.66GHz, 12MB L2 Cache, 1333MHz FSB, 80W)
4GB DDR2-667 Kingston Registered ECC Memory
500GB Western Digital RE2 SATA Hard Drive
Buit-on-Motherboard ATI Graphics
DVD-ROM SATA Drive
Diskette Drive

I just installed the latest build, Alpa 5 of the Jaunty Jackalope (Ubuntu) Linux Distribution, and updated it with all the up to the minute patches, etc.

I am also using the 6.24 beta from Folding@Home for Linux.

Even with using the -smp & -smp 8 flags, I can NOT get 100% usage out of my 8 cores, even with as many as 4 seperate instances of the clients running.

Now, I will be perfectly honest and tell you I do not know that much about Linux, I don't even know how to do a screen grab so I can show you the proof of what all is going on.

But, that's why I'm here. I'd like to learn from all you out there who know Linux, and who know running this software on an 8 core box.

I appreciate any and all help!!!
 
How much less than 100%? From what I've noticed on my setup anywhere between 85-100% seems normal.
 
Ditto, it's known that there are some issues out there, some of the newer kernels (maybe newer than even in your alpha of Jaunty Jackelope) fix it, something with the "completely fair scheduler".
If you check which core you're running, the a2 core is more efficient than the a1, but you never know which one they'll give you on any request. There's further discussion of it over at the project forums here: http://foldingforum.org/viewtopic.php?f=44&t=6007&start=15
FYI, I'm running two SMP instances on my quadcore and it doesn't go to 100% usage, but I do get more PpD than might be expected.
 
Hi lelliot :)

Is this under VMware or native Linux? If in VMs, you will never see 100% CPU usage, it's just the things are in a VM environment. That's the reason some users install a uniprocessor console client in the Windows host to pick up the slack cycles. You can try that to see close to 100% utilization.
 
It's in native Linux, I got rid of the VM's to try & increase my PPD & WU output!
 
Ok, so I finally got two screenshots, and some more information.

First, here is the system monitor output for the 8 CPU cores. Now these come from 8 actual cores, no SMT here:
SysMonitor.png



Second, here is a screenshot of the 4 F@H SMP Clients that are currently running:
FaHSMP.png


Here is what each one is running, they all started about the same time:
fahsmp = 3064
fahsmp2 = 2653
fahsmp3 = 2653
fahsmp4 = 5101
 
Since you're running in native Linux, have you tried using the -smp 8 switch with only one client? There was someone who posted a while back stating that he is running this configuration in native Linux and reported producing about 2000PPD higher output than separate client installs. I can't confirm this myself because I also run the GPU client in Windows alongside VM Linux SMP, but maybe you can give it a try. It should be very easy to test it out by dropping 3 clients and if one client produces more, I wouldn't be overly concerned about full CPU utilization as long as your box is operating at its maximum attainable potential.

Best of luck. :)
 
You must log in or register to reply here.
Back
Top