• 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.

2600k Folding

P

Paragon

Gawd
Joined
Mar 12, 2003
Messages
546
Greetings, All!

Figured I'd drop in! (there is a question at the end)
My old P4 system (I'm talking single core HT here) board crashed a few months ago so I ended up upgrading the system. Remembered about F@H and figured I would run some units for a bit to help the cause.

2600K, TS120
16GB RAM
Win 7 64bit
No video card - iGP

I was running at 4.5GHz for a while then worked on upping it a bit. After a lost WU the other night at 4.8GHz, I found a somewhat stable place at 4.7GHz while keeping voltage around 1.35 to be on the safe side. IBT 4GB stable for 20 runs, always seem to have issues with prime.

I just upgraded to the Client v7 last night (wish I got it in the first place - I like the ability to pause if I need to instead of closing the client) so I finished the WU on the v6 console while having v7 open (Put CPU Priority on v6 and -oneunit) but for some reason, the server found an error on that WU (I think all the shutdowns working on the OC messed it up - Not a good weekend for folding for me..). Good news is the v6 client shut down and the v7 took over the CPU load just fine so right now I am trying to catch up with the v7 WU (2 steps left)

Question: Is it worth running VMPlayer and Ubuntu and fold through that or should I stick with the Win7 client? *I am just folding smp (read the -bigadv will end for anything lower than 16 threads in Jan so might as well help the backlog)

Oh.. I know how you all like photos..
2600K.jpg
 
I gained 17% PPD going from Windows 7 to Native Ubuntu/Linux Mint, gains in VM are similar around 15% Ive heard. Ive got a similar setup to you, I had my 2600k @4.8 but after 2-3 days 24/7 folding I got a lockup, down to 4.7 and has been 24/7 for a few weeks now. Getting around 40,000ppd in Linux.
 
is that with -bigadv? I have VMPlayer 4 installed. Still need to reboot. Will have to find how to install ubuntu.
 
lucky you mine (@ 4.6), only seems to get 30-34 KPPD. When I could get a bigadv WU, it was right around your numbers
 
ReUnitePangea said:
No normal smp it varries between 38-42 depending on WU, has 2133 ram also.
Click to expand...

I wish I had your luck of the draw my 2700k at 4.9Ghz with the ram running at 2260Mhz 10-11-10-30-2T gets between 24k (7011) and 46k (8001) depending on the wu. There are very large variations in PPD when folding smp wu's.
 
jfb9301 said:
as does mine..... see the second rig in my sig.
Click to expand...

Ahh your running windows on that rig, that makes sense, when I had win7 on my rig I got around 32-35k, when I went to linux mint (on that rig, stats were the same when I had ubuntu) I shot up to 38-42k
 
Finally got it up and running in Ubuntu. Used VMware 3 and the newest version of Ubuntu. Originally forgot the "8" with the -smp and was wondering why the times were really large (about 14 minutes). Also had duplicate step readings

-smp -verbosity 9
Code: 
[03:24:11] Entering M.D.
[03:24:17] Mapping NT from 8 to 8 
[03:24:24] Completed 0 out of 500000 steps  (0%)
[03:30:12] Completed 5000 out of 500000 steps  (1%)
[03:40:46] Completed 5000 out of 500000 steps  (1%)
[03:47:03] Completed 10000 out of 500000 steps  (2%)
[03:58:05] Completed 10000 out of 500000 steps  (2%)
[04:03:47] Completed 15000 out of 500000 steps  (3%)
[04:14:34] Completed 15000 out of 500000 steps  (3%)
[04:20:09] Completed 20000 out of 500000 steps  (4%)
[04:30:39] Completed 20000 out of 500000 steps  (4%)
[04:36:15] Completed 25000 out of 500000 steps  (5%)
[04:46:46] Completed 25000 out of 500000 steps  (5%)
[04:52:35] Completed 30000 out of 500000 steps  (6%)
[05:03:02] Completed 30000 out of 500000 steps  (6%)
[05:09:00] Completed 35000 out of 500000 steps  (7%)
That was only seen directly from the log. The terminal window looked normal. Just with the large TPF times.

-smp 8 -verbosity 9
Code: 
05:38:22] Mapping NT from 8 to 8 
[05:38:22] Resuming from checkpoint
[05:38:22] Verified work/wudata_01.log
[05:38:22] Verified work/wudata_01.trr
[05:38:22] Verified work/wudata_01.edr
[05:38:22] Completed 50182 out of 500000 steps  (10%)
[05:40:44] Completed 55000 out of 500000 steps  (11%)
[05:43:10] Completed 60000 out of 500000 steps  (12%)
[05:46:07] Completed 65000 out of 500000 steps  (13%)

I'll let it cook overnight and see how it goes. Currently have the -bigadv on to see if I could get one to check speed.
 
Last edited:
Youre times are the same. smp = smp 8 on a 8 core/thread machine.

smp hates to be intrupted, so if youre running day to day stuff, youre going to always get variation on the tpf.
 
Patonb said:
Youre times are the same. smp = smp 8 on a 8 core/thread machine.

smp hates to be intrupted, so if youre running day to day stuff, youre going to always get variation on the tpf.
Click to expand...

The initial step is about the same but after that, there wasn't much going on to induce a change like that (mainly web). I don't know what voodoo was going on to take so long per step and what is with the duplicate steps?

Code: 
[04:20:09] Completed 20000 out of 500000 steps  (4%)
[04:30:39] Completed 20000 out of 500000 steps  (4%)

Could it have been just that it was the first run? baffling indeed but I did try the -smp and it was running fine now. :confused:
 
Web browsing can make a huge difference to TPF's, especially for flash heavy sites etc, I usually add 20 seconds per frame during daily use on my rig for a normal SMP project
 
Typical browsing seems to add about 10-15 seconds. I'm sure youtube would increase about double that.

OK.. so I nudged the CPU up to 4.8GHz. Offset is +0.110. IBT will cause the voltage to go to 1.404. Folding it holds at a nice 1.368. The problem I am seeing is that the CPU load is no longer a constant 100%. Threads will drop to 99.3 and back up to 100% and I think it is increasing the possible TPF slightly. When I first installed VMWare, I noticed once I loaded the VM, the load would be at 100% full time even without F@H running (Full load but not a heavy load so temps were mild) Now I did move the VB from the SSD to the spinner. I'm wondering if during the reconfiguration, something changed to cause the load drops but that was just the Box folder that I moved, not the player.
priority.ungrabbed = "idle" is still in the config file and I tried adding priority.grabbed = "idle" also. Threads still drop down.

Any ideas? Do I need to reinstall VMWare? Running v3.0.0

*Same thing when I dropped back down to 4.7GHz. I do see "NOTE: Turning on dynamic load balancing" now when folding starts.
 
Last edited:
Updated to F11F beta BIOS and was able to get 8 runs of IBT 4GB before it would lock up (would lock up on the 2nd before) so I upped the voltage to 0.80 offset (1.344-1.356 under load) and it ran the 20 just fine. I think I may drop the clock to 4.5 and let the voltage get below 1.3 as I don't want to have to deal with a dead CPU anytime soon.
 
Yah well mine (2600K) wouldnt take above 4.5ghz. I had it at 4.7 for a while but it would crash every now and then. So I dropped it little by little till I got to 4.5. She has been stable for several months at that speed. My voltage is high, 1.416 but I haven't fooled with it. Left the automatic stuff do its thing as I don't know a lot about it. Tried playing with it on manual but crashed a few times. Since it went stable at 1.416 to 1.424 as reported by CPU-Z and my temps are good I have just left it alone. This is all under Windows 7 Ultimate 64bit on a Sabertooth P67 board.
Just had to reboot system to get the system software to run. When I rebooted it went to 4.7ghz again. Guess I never saved the settings for 4.5. Anyways she is running now at 4.7 and I think I will let her run and see what happens. Temps for the CPUs are 73-74C.
FYI


I get around 33K ppd on most WU. Some of them drop to the 28K region but I don't remember which ones. And yes am running SMP 8 and of course bigadv.
 
Well that didn't take long. She crashed so have moved it back to 4.5ghz. Temps and voltages pretty much the same.
 
That voltage seems high for automatic. I can leave mine on automatic for 4.5 and the voltage under load is 1.272 (HWInfo) At 4.7 I needed to up the voltage for stability. Using VM Linux, it is right around 32k PPD (these are all SMP units, have yet to get a bigadv) and with it at 4.8 was around 35k PPD. Current TPF is around 2:15. It ware 2:07-2:09 at 4.8.

I may let it run at 4.7 and see how it is for stability.

OK.. One thign I have noticed from the start.. The clock in windows seems to lag behind. Have not noticed if it is only under load but at one point when I was folding through the day, there was a 10 minute difference when I did a manual update on the internet time. Just did it and it was 2 minutes back. I don't thing the VM Linux has that issue. What would cause that?
 
Paragon said:
OK.. One thign I have noticed from the start.. The clock in windows seems to lag behind. Have not noticed if it is only under load but at one point when I was folding through the day, there was a 10 minute difference when I did a manual update on the internet time. Just did it and it was 2 minutes back. I don't thing the VM Linux has that issue. What would cause that?
Click to expand...

I remember troubles with this years ago using the notfred image. And now my big machine is doing it (it wasn't before). My 970s linux VM is consistently loosing time, 20+ minutes over the last several days. I'm using the linuxforge image and I have ntpd installed. The image does not support the service command so while I can kill ntpd and manual update, nothing I have found on the web allows me to restart ntpd.
 
Just an update. Bumped the core and RAM voltage a bit. 4.7GHz 1.356V on the core (+0.1 offset) and the RAM is sitting around 1.53V. Almost 4 days straight WCG tasking all 8 threads 100% and we are still running. I think I could actually get to 4.8 or higher stable as the RAM voltage helped a ton however the voltage would have to be 1.4+ so I'm not sure if I want to push it that much.
 
I have a 2600K that I have left the voltages on automatic. Reporting from CPUZ they say it is at 1.416 and fluctuates to 1.424. This has been runing at 4.5ghz for several months with no problems. Temps in the 70 to 72 C on air. It will NOT run at 4.6 or 4.7 for more than a half hour but is fine at 4.5 I am afraid to crank up the voltage any more so haven't tried to go higher.

FYI
 
I would check to make sure you have the latest version of CPUZ or try another program. Initially, CPUZ was not reading my voltage properly (The most recent version does) and CPUZ HWMonitor was not reading the fan RPM correctly. I end up using HWInfo for all my info. It really depends on the motherboard as to how well a program will read everything.

On auto, at 4.5, the VID should be around 1.386 so unless you have LLC enabled, you should be running around 1.27 V at that speed. Mine works fine on auto voltage with the voltage just stated. Have not run it at 4.6 for an extended time on auto but voltage was just shy of 1.3
Around 1.42, you should be able to run at 4.8 or 4.9 with some tweaks.
 
Yah well CPUZ agrees with the Asus motherboard program so I think the voltages are correct.
I once tried to play with the voltages and it kept crashing. Switched back to auto and it would run for long periods. Havent messed with it since.
 
You must log in or register to reply here.
Back
Top