Go Back   Hardware Canucks > HARDWARE CANUCKS COMMUNITY > HardwareCanucks F@H Team

    
Reply
 
LinkBack Thread Tools Display Modes
  #21 (permalink)  
Old October 6, 2010, 06:06 PM
bkh2's Avatar
MVP
F@H
 
Join Date: Jul 2009
Location: Coquitlam, BC
Posts: 265

My System Specs

Default

Quote:
Originally Posted by Charloz24 View Post
Try also just running with -smp 12. My i7 980x fold bigadv for more than 6 months perfectly, But for some unknown reasons, if I try -smp 11 it error on all non bigadv unit. With -smp 12, everythings runs fine.
Will do. I'll probably usually use -smp 10 since I use my machine to play games while folding. Just for reference, could you tell us your PPD using 12 threads, assuming you're folding bigadvs, in addition to the frequency of your CPU? Would you happen to have numbers with, say, 10 threads and smp versus smp with bigadv flag?

On a side note, how well do ATI cards do in folding, more specifically the 5870? I read that folding requires CUDA, but there are ways to get around that. I also read that ATI cards do not produce as many points, assuming it was able to fold. However, I was unable to find numerical comparisons. Would anyone happen to have values they'd like to share?
Reply With Quote
  #22 (permalink)  
Old October 6, 2010, 06:51 PM
Sagath's Avatar
Hall Of Fame
F@H
 
Join Date: Feb 2009
Location: Petawawa, ON
Posts: 2,524

My System Specs

Default

There is a few bad regular A3 WU's floating around. Ive hit 5 or 6 in the last two days. Really killing my PPD :( And no, its not my OC being unstable, its the WU's. Ive been folding -bigadv's A3's just fine. Check out the F@H forums for more info about it.

As to ATI Folding, the PPD sucks. I dont have hard numbers, but they are out there if you google it. To be honest, IMO its not worth the hydro bill for the amount of PPD you get.
__________________
Quote:
Originally Posted by Lpfan4ever View Post
AKG shot the hamsters, and then Perineum drove his Mustang with summer tires into the server.
My Disclaimer to any advice or comment I make;
Quote:
Originally Posted by CroSsFiRe2009 View Post
I'm a self certified whizbang repair technician with 20 years of professional bullshit so I don't know what I'm talking about
Reply With Quote
  #23 (permalink)  
Old October 7, 2010, 04:16 AM
Aix's Avatar
Aix Aix is offline
Top Prospect
 
Join Date: Oct 2008
Location: Mississauga, ON
Posts: 91

My System Specs

Default

Quote:
Originally Posted by bkh2 View Post
A copy of my log:

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

[21:33:53] Folding@home Core Shutdown: EARLY_UNIT_END
[21:33:56] CoreStatus = 72 (114)
[21:33:56] Sending work to server
[21:33:56] Project: 6062 (Run 1, Clone 52, Gen 30)
The bolded text is what you absolutely don't want to see and is the result of a bad -smp #. There's a thread on the official F@H forum (Folding Forum • View topic - 15 "Instant" EUEs in the last 30 hours) that talks about this issue.

And yes, EUE or Early_Unit_End's result in you prematurely returning WU's to Stanford, meaning they will affect your bonus points potential if they put you under the 80% completion qualifier.
Reply With Quote
  #24 (permalink)  
Old October 7, 2010, 08:22 AM
rjbarker's Avatar
Hall Of Fame
 
Join Date: Feb 2008
Location: Courtenay, B.C
Posts: 5,932

My System Specs

Default

bkh2 (not to step on anyones folding toes)

Just a heads up with OC'ing your 980X......LinX and IBT are not the optimal Program to check for stability with these Chips (very well documented and I can attest to it through a ton of testing).....passing either of those tests doesn't mean a whole lot.....yes your "stable", but I doubt your stable enough to Fold.
For the 980X run:
P95 Small FFT for at least 6 hrs and
P95 Blend for at least 6 hrs

vCore / QPI (vtt) had to further bumped up well beyond what was "stable" running LinX / IBT to get stable P95.

I consistently found (on every occasion early in my OC'ing) that passing LinX Max Memory (All) woukd result in a BSOD running P95.....we're basically waiting for a LinX / IBT update to address the 6 Core Processor.
btw..this is specific to the Six Core CPU.....I found the opposite holds true for Quad Core....
__________________
Introducing me n my OCD to Watercooling, is like taking an Alcoholic to an "all you can drink" Beach Bar in Mexico

.
Reply With Quote
  #25 (permalink)  
Old October 7, 2010, 09:33 AM
Aix's Avatar
Aix Aix is offline
Top Prospect
 
Join Date: Oct 2008
Location: Mississauga, ON
Posts: 91

My System Specs

Default

Quote:
Originally Posted by rjbarker View Post
bkh2 (not to step on anyones folding toes)

Just a heads up with OC'ing your 980X......LinX and IBT are not the optimal Program to check for stability with these Chips (very well documented and I can attest to it through a ton of testing).....passing either of those tests doesn't mean a whole lot.....yes your "stable", but I doubt your stable enough to Fold.
For the 980X run:
P95 Small FFT for at least 6 hrs and
P95 Blend for at least 6 hrs

vCore / QPI (vtt) had to further bumped up well beyond what was "stable" running LinX / IBT to get stable P95.

I consistently found (on every occasion early in my OC'ing) that passing LinX Max Memory (All) woukd result in a BSOD running P95.....we're basically waiting for a LinX / IBT update to address the 6 Core Processor.
btw..this is specific to the Six Core CPU.....I found the opposite holds true for Quad Core....
I found this to be the case as well; I can leave LinX going for 100 runs at max RAM and get no errors, then leave Prime95 going for 9 hours and have a worker thread fail (not BSOD).

I honestly just don't have the time or patience to mess around with Prime95 anymore and I've found that being LinX stable is good enough for F@H (*knocks on wood*).

Edit: Oh, and from HardOCP MSI Big Bang motherboard review (Overclocking & Conclusion - MSI Big Bang XPower Motherboard Review | [H]ard|OCP):

Quote:
For my manual results I did the normal thing of turning off all CPU features and Spread Spectrum, but leaving on HT and all cores of course. I set all voltages to AUTO except for CPU and RAM. I bumped the memory to 1.75v and left it there. For the CPU I left it set to 1.25v and started scaling up the BCLK moving the multiplier down to keep the processor around a 3.5GHz clock speed. Once I found my upper end bus of 221MHz I then started moving the processor multiplier up. The 4.2GHz I am showing you below was not 100% stable. Under Prime95 stress testing, sometimes one of the threads would fail. You could pretty much do anything with the computer expect stress test it though. Encoding video was not issue either. Even when Prime failed on a thread I could load that thread with another stress program and keep it running at 100%. Maybe time for us to start looking beyond Prime95, but we have felt that it has generated some loads that seemed more "real world" to us than things like linpack.
Reply With Quote
  #26 (permalink)  
Old October 7, 2010, 09:52 AM
rjbarker's Avatar
Hall Of Fame
 
Join Date: Feb 2008
Location: Courtenay, B.C
Posts: 5,932

My System Specs

Default

^^^ Yeah I found this extremely frustrating (P95 Tests)......likely why so many were initially posting such low voltage (running Linx) requirements for fairly moderate OC's (as in 4+ Ghz)......

Cheers
__________________
Introducing me n my OCD to Watercooling, is like taking an Alcoholic to an "all you can drink" Beach Bar in Mexico

.
Reply With Quote
  #27 (permalink)  
Old October 7, 2010, 11:13 AM
bkh2's Avatar
MVP
F@H
 
Join Date: Jul 2009
Location: Coquitlam, BC
Posts: 265

My System Specs

Default

Normally, I would always run P95 but so many people have been saying IBT was much better. Unfortunately for me, I guess, was that they were talking about quads and under. So far, my folding logs do not indicate that any EUE's were present after I changed to -smp 12. About the 80% completion qualifying thing, if I apparently sent back 27 bad units, would this mean I need to send back 135 good units (27/135 = 20%) before I get any bonuses? Would you recommend I stress with P95 before continuing folding, or should I be fine at these settings?
Reply With Quote
  #28 (permalink)  
Old October 7, 2010, 11:37 AM
Aix's Avatar
Aix Aix is offline
Top Prospect
 
Join Date: Oct 2008
Location: Mississauga, ON
Posts: 91

My System Specs

Default

Quote:
Originally Posted by bkh2 View Post
Normally, I would always run P95 but so many people have been saying IBT was much better. Unfortunately for me, I guess, was that they were talking about quads and under. So far, my folding logs do not indicate that any EUE's were present after I changed to -smp 12. About the 80% completion qualifying thing, if I apparently sent back 27 bad units, would this mean I need to send back 135 good units (27/135 = 20%) before I get any bonuses? Would you recommend I stress with P95 before continuing folding, or should I be fine at these settings?
Unfortunately yes - as I understand it - you will have to build your ratio back up by doing a week or two of SMP without qualifying for bonus points.

It doesn't sound like your overclock was the problem (that usually results in Unstable_Machine errors) but it was the -smp 11 flag that you used, which is unfortunate. I have an ATI card (5870 when I tried it) and when I ran the GPU client I found that the PPD from the 5870 was massively hurting my -bigadv PPD, so I just canned it right there. I suppose your GTX 275 would provide a much better PPD offset than a 5870 though. I never bothered messing around with anything other than -smp 12.
Reply With Quote
Reply


Thread Tools
Display Modes

Similar Threads
Thread Thread Starter Forum Replies Last Post
How To: VMware & Bigadv Folding geokilla HardwareCanucks F@H Team 10 December 1, 2012 09:56 PM
Turned off my i7 bigadv folding machines Rison HardwareCanucks F@H Team 2 June 29, 2010 11:01 AM
Several noob questions GuelphLen HTPC Systems & Software 4 May 5, 2010 07:44 AM
bigadv folding help...?! ADay2Long HardwareCanucks F@H Team 12 March 16, 2010 04:06 PM
Noob Questions Hunt HardwareCanucks F@H Team 7 January 29, 2010 08:29 AM