OK first finding the process ID.
 
from a blank console type top.
 
        
        
 
You will get 
 

That will show you your process ID's. press q to stop it and go back to your command prompt.

Now this should work I just did it on mine. Note this is my machine and will not work for you

 as written you need your PID. 

sudo taskset -pc 1,2,3,4,5,6,7,8,9,10 1230        
        pid 1230's current affinity list: 0-7

Broken down its 

sudo taskset -p= process c= cpu 1,2,3,4,5,6,7,8,9,10=cpu numbers 1230=Process ID

That should make PID 1230 only run on cpu 1-10 leaving CPU 0 and 11 alone
Then I set cpu 0 and 11 as gpu folding as shown.
 

 
*edit of course I did not notice the failure on the second gpu core
 
***Edit.. light bulb goes on....  of COURSE it failed.. I only have 8 cores in this cpu its my 3770 ...
 
You can do a mask too like 
  taskset -pc 1-6 1230   

jini, you make that look so easy, I suspect that if I had the ability (hardware) and the need then I could do that. lol

Locking core 17 to a cpu core in Linux didn't help while smp folding, the TPF actually got  worse.
Lock to a core: 3:16 7663 (6, 24, 31)

05:23:00:WU00:FS01:0x17:Completed 1825000 out of 2500000 steps (73%)  
 05:26:16:WU00:FS01:0x17:Completed 1850000 out of 2500000 steps (74%)

Running default: 2:41 7663 (2, 41, 14)

07:20:07:WU01:FS01:0x17:Completed 325000 out of 2500000 steps (13%)  
 07:22:48:WU01:FS01:0x17:Completed 350000 out of 2500000 steps (14%)

 
Two different WU but both 7663 so I don't think it being a different clone would make that much of a difference.
Diffidently doesn't work out the same as it does in W7 when locking to a core.
Seems there still should be some way to get some improvement while smp folding
after all there are 4 free cores for core 17 to pick from. If there is no fix I can't see
where it be worth GPU folding in Linux if you also plain to smp fold at this time.
Maybe further down the road it'll get better.
 
Edit: Forgot to mention both test were done in pure terminal with the GUI killed
which did seem to improve the TPF a little over running it in the GUI.

Quisarious

TheWolf If there is no fix I can't see
where it be worth GPU folding in Linux if you also plain to smp fold at this time.
Maybe further down the road it'll get better.

I am folding with 4 gpus and SMP 28 (32 cores avail) in linux without problems.  I'f you'd like to jump on IRC at some point, maybe together we can troubleshoot your issue?

Thanks for the offer Q, might look into that.
What is strange is this morning and the last WU I worked last night
both seem to be doing a frame at around 2:15. Not sure why all of a
sudden a improvement other than I see with both WU's core 17 ended
up picking cpu core 20 to land on. Before it might land on any of the other
3 cores and TPF would be different which each. GPU folding only I was seeing
2:04 TPF so at least its much closer all of a sudden now. I'll keep a eye on
it to see if it changes CPU cores and if it goes back to the longer TPF.
Still running in pure terminal but have the GUI active to monitor TPF.
 
Scratch that, while sitting here typing the TPF has jumped back to 3:34
still on cpu core 20. Only thought that comes to mind is the TPF maybe
improving after the 10min screen mointor is shut down. Now 4:00 TPF.
I using another computer to post from so no extra activity going on , on
that rig. I'll let it complete a few more WU's so I can try to find out why
the TPF is changing so much. Mointor just went off, so I'll wait about
30mins and check the TPF then.

jpeck2009

Just checking, but is anyone still pulling the new beta units?  My titan has pulled a couple of non-beta old 0x15 core units in a row now. 

No problems here pulled 3 in the past hour. 2 just a moment ago.

jpeck2009

Just checking, but is anyone still pulling the new beta units?  My titan has pulled a couple of non-beta old 0x15 core units in a row now. 

 
1 of my 5 gpus pulled a 7426 lastnight out of nowhere. its been nothing but 7663 the rest of the time.

One my three GPUs pulled a slow Core 15 7625 last night, but that's the only non-Core 17 7663 I have seen in several days.
 

Man that sux!
I'm begining to think we are out running the core 17 project supply with so many doing these.

I'm battling 8101's on both 4ps. Man I hate those.
 

Well would you lookie there! And that's without my 2x480 & 660ti.

the new core17 wu is carrying me good. busted into the top 100 last night and 40million. and puts me 6th on the team last update.

TheWolf

Well would you lookie there! And that's without my 2x480 & 660ti.

The numbers you put up are awesome! Keep up the great folding Wolf!

HFM only knows what the current WU's would add up to at the current TPFs if worked at those rates for 24 hrs straight so once a WU is finished and a different WU is downloaded or a TPF changes the old estimate is obsolete.
 
The 6172's finally finished their 8101 and got an 8103, yay! The 8439's won't finish their 8101 for another 4 hours. Five WU's working around the clock and I got a goose egg for the 9am update! The 680's with their 1:35 TPF on the 7663's apparently perfectly straddled the update.
 
I got the first coat of black paint on the basement folding rack and will hopefully get the second on later today so I can move the 4p's into it tomorrow and then get cracking on the 4p ankle biter 8425's, for a welcome PPD boost.
 

Ahh, that's old news

"Jesse_V"Proteneer just restarted the Work Server and uploaded a new version of the core.
<Jesse_V> what changes are in the new version?
<proteneer> not much on the performance side
<proteneer> its just a minor patch to make checkpointing more robust