Sign in with
Sign up | Sign in
Your question
Solved

Best Stability Test with limited time

Last response: in Overclocking
Share
July 13, 2011 9:50:02 PM

Hi All,

I'm a rookie OC'er and I'm wondering what the pro's opinions are on stability testing. I just read this guide and I like how helpful it was. In summary, it basically says for optimal and perfect testing to use Prime95 on Blend for 24hours, 3dMark for 4 hours(max), and memtest86 for 24hours.

Unfortunately I don't have this kind of time, and I was hoping that experienced members have suggestions on the most "efficient" way to stress test my basic cpu overclock with limited time.

Right now I'm going for large-FFT testing to test both my temps and stability at the same time.

Rig:

i5-2500k (3.3Ghz)
MSI P67-G45
Gelid Tranquillo Cooler
8Gb (2x4Gb) Mushkin PC3-10700 DDR3 memory

Latest "stable" (ran Prime Large-FFT's for 15 minutes) was an OC to 4.0GHz (x40 multiplier) with a 1.15V Vcore, temps max at 64C*C under full load.

Suggestions for efficient stress testing with limited time?

Best solution

a b K Overclocking
July 13, 2011 11:35:58 PM

Honestly, there is no magic number of hours (or minutes) that will determine a perfect OC. What you read is a guide, not a requirment; however, longer testing durations with stability will prove itself in the long run. Will running P95 for 24 hours, stable, guarantee that your system will always be stable? No, but you can be a lot more confident that your system won't fail on you.

The beauty of P95 is that it creates an error log, of sorts. This file is called "Stress.txt" and it will show you what you need to know about why your system crashed. That said, if you don't have time because you're just very anxious to use your new toy, then by all means skip the testing now, but allow the test to run when you can step away from the computer for awhile; while you're at work/school, for example.

Should you employ this idea, and you come home to a crashed system, just get into BIOS and load defaults. Next, check the Stress.txt file for info on the crash.

Share
July 14, 2011 1:32:13 AM

^ +1

Did I read your post right?

1.5v for 40x multi? That's pretty excessive. You should be able to do a 40x multi with stock voltage (1.2v). 1.5v is for more extreme OCs (50x multi). And I don't think that's a safe voltage and may reduce the life of your cpu.

I can use 1.195v @ 4.0GHz and be P95 stable over a 7 hour run. I recently reduced voltage to 1.190v and ran stable for 3 hours. I generally run tests overnight when I sleep. It also helps the electrical bill as this will be running on the cheapest tier.
m
0
l
Related resources
July 14, 2011 3:56:43 AM

neograndizer said:
^ +1

Did I read your post right?

1.5v for 40x multi? That's pretty excessive. You should be able to do a 40x multi with stock voltage (1.2v). 1.5v is for more extreme OCs (50x multi). And I don't think that's a safe voltage and may reduce the life of your cpu.

I can use 1.195v @ 4.0GHz and be P95 stable over a 7 hour run. I recently reduced voltage to 1.190v and ran stable for 3 hours. I generally run tests overnight when I sleep. It also helps the electrical bill as this will be running on the cheapest tier.


WHOOPS, meant 1.15V, now cooking x41 @ 1.7 Ghz...
m
0
l
July 14, 2011 3:20:45 PM

CookieMaker said:
WHOOPS, meant 1.15V, now cooking x41 @ 1.7 Ghz...


Wow, I envy your CPU. Mine becomes unstable at 1.185v @ 4.0GHz (within 30 mins of a blend test). It's 'stable' at 1.190v with a blend test running for ~7-8 hours.

I think to really check stress/stability, you really need to run long tests. Some users have reported errors when running it a little past an hour (which shows some instability). I can boot to BIOS and load Win7 x64 Ultimate fine on my comp at 1.185v, but it was still an unstable OC under testing. Yes, it passed the 15 min mark without errors, but within a half hour, it failed.

I'm leaving my OC @ 4.0GHz. It gives me the voltage (< 1.2v) and the temps (< 55'c) I want. So I'm quite happy with it right now. I don't really think I need any more juice than that. I don't think I'll see any huge performance gain if I clocked 100-500MHz faster getting higher volts and temps.
m
0
l
a b K Overclocking
July 14, 2011 3:56:48 PM

Wow. Default CPU voltage is 1.2v, so you guys are really lucky getting those overclocks at those low voltages.

Cookie:
You do know that you can run your normal programs while Prime95 is going in the background right? Even if you're gaming, your CPU won't be fully loaded and Prime95 will use those extra CPU cycles for its testing. Of course, you can't do anything while 3DMark or MemTest are running though.

If Prime95 crashes, even on just one worker, you can assume the voltage isn't enough. Raise it just a bit and try again.
m
0
l
July 15, 2011 12:37:46 AM

Leaps-from-Shadows said:
Wow. Default CPU voltage is 1.2v, so you guys are really lucky getting those overclocks at those low voltages.

Cookie:
You do know that you can run your normal programs while Prime95 is going in the background right? Even if you're gaming, your CPU won't be fully loaded and Prime95 will use those extra CPU cycles for its testing. Of course, you can't do anything while 3DMark or MemTest are running though.

If Prime95 crashes, even on just one worker, you can assume the voltage isn't enough. Raise it just a bit and try again.



Really?? No I definitely NOT know that, i thought that for accurate results you had to run it solo. I'm assuming any gaming you do while running it though is extremely laggy.

Just tried out x44 (4.4GHz) on 1.26V and completed 5 ITB's (hit 89*C at hottest) but failed Prime after 2 hours :( 

I haven't changed any memory timings (except forced them to run at 1600, which they are supposed to) or voltages but i'm going to run memtest anyways to attempt to diagnose the prime failure (hoping i don't need more vcore).
m
0
l
July 15, 2011 1:13:17 AM

I tried to OC to 43x multi and found I needed on avg 1.289v with max temps hitting 65'c on cores 2/3 and lower (62/61'c) on cores 1/4. This ran P95 blend for ~8 hours with no errors. While 45x multi required > 1.3v with peak temps in the mid-high 70s.

Too high for my tastes. So I cranked back down to my happy medium.

Cookie, even if you got 45x multi stable, you really have to lower those temps. They're too high. Temps should be kept to 70'c. Or that's to my understanding.
m
0
l
July 15, 2011 1:19:58 AM

I'm trying, most of the time on full blast they're just above 70*C (which I've seem some people say they live at) but I'm also concerned. I'm going to reinstall the heatsink and hope for better temps. I found that it was running really hot w/o any OC when i first fired it up on the stock cooler, i was still in the mid-70s under prime load, maybe i just have a hot chip or mobo :( 
What cooler are you using? My Tranquillo is supposed to be almost as good as the Hyper 212 but I've never seen it running cooler than like 38*C idle.
m
0
l
July 15, 2011 5:03:01 AM

I'm using a Scythe Mugen 2 Rev.B. I made a newb mistake when installing it the first time; I didn't secure it properly. Temps were high at stock performance even after a CMOS reset (idle around 38-40'c and 70'c with P95).

But after reseating it, temps shot straight down. At idle, my temps vary from 30-35'c. Load in P95 at 40x doesn't exceed 55'c and at 1.190v. I have to admit, the second time around, it wasn't as daunting to install the huge heatsink. But it took some time as I patiently cleaned off the old TIM and applied a new one.

Idle temps don't really matter, but the load temps do. The temps you live with will be up to you, though it's suggested to keep it below 70'c. My OC preference was to see how far I can push at stock voltage (< 1.22v) and still keep a low temp (< 60'c). I wasn't really aiming for any multi. Just good voltage and temps. :p 

The farthest I've been able to go was 40x multi. Can't really complain about that accomplishment. I'm quite pleased with it. Should last me a long while... :D 

Try to reseat the heatsink and see if there's any improvement. If not, you should be able to keep it at 40x multi. What's your ambient temp? I think that makes a difference as well.
m
0
l
July 16, 2011 1:05:40 AM

Dunno really what my ambient temps are, cpuz and coretemp dont show them... I'm going to reseat the heatsink in a vertical fashion, right now i think the air circulation isn't perfect (and I also believe i used too much thermal paste, gotta check to be sure). Hope for the best.
m
0
l
July 16, 2011 4:03:18 AM

Ambient temp is your room temp. ;) 

I hope everything turns out fine in the end.
m
0
l
July 23, 2011 2:53:09 AM

Best answer selected by CookieMaker.
m
0
l
!