folding and OC stability

rangor gubbins

rangor gubbins

rangor gubbins

Associate
Joined
5 Aug 2006
Posts
1,117
Location
Kent, UK
What are people's opinions on length of time to run ORTHOS stability check for a PC that is folding (F@H)?

At my current OC (11x267) i stopped the small FFTs test after 12hrs with no problems.
But running the blend test ORTHOS crashed after 19 hrs. I don't whether the program crashing is a more serious sign of instability than just an error in the calculations.

Temps are 25 idle, 56 load
 
Depends a bit on which client you're running. The linux smp one is definitely more savage than orthos, it seems to be more aggressive than ibt as well. At least I'm stable during ibt but unstable during folding.

I'd also be inclined to just run folding, the code is pretty good at spotting unstable computers. If you're getting units thrown back at you, you're not stable enough. However I'd want to be stable for many runs of ibt before trying this.
 
F@H is a stability test all by itself. I've had things pass 24hrs of Orthos and fall over with a decent WU.

In some ways F@H is the ultimate stability test because you're maxing out all your CPU and GPU cycles all the time.
 
LinX with max available memory is the best stability test I've come across.

A fully stable CPU should be able to run any test without failure though, you don't just pick and choose which suits.
 
Last edited:
rangor gubbins said:
At my current OC (11x267) i stopped the small FFTs test after 12hrs with no problems.
But running the blend test ORTHOS crashed after 19 hrs.

Temps are 25 idle, 56 load

Intel® Pentium® E2200 2.20GHz (11x200) @ 2.93GHz (11x267)
Click to expand...
Hello rangor gubbins,

I could use that name for one of my D&D characters, a Hobbit/Ranger, "Rangor Gubbins at your service!" :D

Respect to you sir for a mighty 24 hour blend test attempt, don't see that many folk taking time out these days for proper stability testing that like. Even with the advent of newer testing apps like Intel Burn Test and LinX I think Ye Old 24 Hour blend is still a relevant test . . . The fact you have come so close to passing means that total stability could be just a flick of the wrist away! :)

Some people may tell you to ignore prime, or just use Folding@home as the stability test but I'm not one of those people, I think its important to get the stability testing done and dusted and out the way and then get on with the things you really enjoy, some people acquire a certain Peace Of Mind after their computer system has passed rigorous stability checks that not everyone appreciates, a saying that springs to mind is "Everybody wants to go to the party but no ones wants to stay behind after to clean up the mess!" ;)

If you wanna hand getting your system bomb proof stable I'm happy to help, seems like you are very close indeed, I'm sure that Prime Blend error is buggin the **** out of you heh! :o

If you can mimic this screenshot we can put that overclock of yours under the microscope and see what the problem may be! :cool:

CPU-z

Memset

Core Temp

Up to you and I won't be offended in the least if you decide to wimp out and just be happy with Folding stability! . . .

P.S You can open multiple copys of CPU-z from the same exe, in case you were wondering . . .
 
Last edited:
hi B.W. sorry, i've only just seen this post!! Wish i'd seen it earlier as it would have saved me a lot time over the weekend having some advice.

Anyway, I've done lots more tests. The CPU is definitely stable with 1.45v (1.44-42v speedfan) at 2940Mhz as i've run a small fft test for over 30hrs. I think my RAM timings were a bit tight for the new RAM i put in, so i slackened them off to 5-5-5-15 and re-ran the test. Pass with flying colours (24hrs+ blend test).
Also been testing to see how far the FSB will go. Seems fairly stable at least up to 1500mhz (poss further). I realised that this MSI Neo2 Fr board is actually meant to go up to 1333mhz and i've read of people getting it to 2000. I did find that if i put the Bus speed above 300 the board auto sets the NB to 1.5v and the 'SB I/O power' to 1.8v, which seems like quite a hike from the stock (1.25v and 1.5 respectively).

Memtested RAM at 450mhz for 12hrs. 2.10v 5-5-5-15. passed no errors.

Currently testing these settings:
366x8 = 2928 Mhz with 1.45v
FSB = 1464
RAM is at 439 (5-5-5-15) 2.10v. Do you think a touch more voltage would allow for tighter timings? does it work like that or does more volts just allow for higher frequency?
Testing with LinX, max memory, 20 runs. Its done 5 runs so far with no problems. Temps have maxed at 59 on both cores (the CPU and Tuniq Tower are lapped and fan is running at a very quiet 75%).

I know from my last OC attempt the CPU is stable up to about 3.1Ghz but it needs more volts and gets rather toasty. Would be nice to get it to 3Ghz with the current voltage but i don't really need the extra mhz, it would just be for satisfaction :)

Have downloaded your links and will post a screenie
 
20091222366x8linx.jpg
 
Last edited:
It appears you are using an memory mutliplier to clock your DDR2-800 ram above its rated 400MHz frequency . . . looks like you have it at 439MHz (DDR2-878).

Slackening the timings and raising the vDimm certainly can work wonders with a memory overclock but how far it will go will vary between different sets of memory.

In your boots I would be doing some *isolated* testing on the three key areas . . . . CPU . . . .FSB . . . . MEM

Normally when clocking your chip the first time its advisable to change the memory so its running [1:1] sync, that will remove it from the equation and in theory allow you to clock the FSB to 400MHz and the chip as far as possible before the memory starts freaking out . . .

If you sync the memory to the FSB you will be able to concentrate on getting the CPU stable, once you worked out how much vCore it needs you can then move onto the FSB and memory . . .

Are you cool with changing the ram multiplier to sync [1:1] :)
 
I have, as you have suggested, tested each component individually over the weekend.
CPU - 2940Mhz with 1.45v. (using Orthos small FFT +24hrs)
RAM - tested up to 450 mhz, 5-5-5-15, 2.10v (using memtest 12hrs)
FSB - went up to 1500Mhz (large FFTs orthos test for 30min)

So my current settings of 8x366 and the RAM/FSB ratio of 5:6 means that all the components are slightly under what i tested them at individually. 2hrs of LinX and its not failed yet.
 
I'd say that Folding's as stressful as Prime, but probably less so than the newer stability tests like LinX. Folding and Prime get my CPU to about the same temperature, but IBT and LinX make it a good 10-15c hotter... that alone makes it worth running a stability test. If your CPU can run the above two for a few hours than it shouldn't have any problems with Folding.
 
You must log in or register to reply here.
Back
Top Bottom