• 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.

What to use to test overclocking?

L

Lucasta

Weaksauce
Joined
Nov 22, 2007
Messages
79
Just wondering what you guys use to check your overclocking stability? I've been using Prime 95 which has run for 12 hours with no problems. Im running an I7 920 at 4ghz with 1.25v. Temps seem ok as they run around 65C under 100% load and around 37-40C idle. Is there anything to use to test with? Also what do you use to check temps with. I've been using Core Temp. Just wondering. Thanks guys :)
 
In no particular order:

1. Intel Burn Test
2. LinPack
3. OCCT
4. Prime95
5. Memtest (for RAM)
 
Doesn't OCCT now include a LinPack-like test? I like running it's GPU tests alongside Prime95 to test overall stability after I've stress-tested the individual OCs on GPU, CPU, and memory (w/Prime, OCCT, Memtest, etc). I use SpeedFan to monitor overall temps but I verify my CPU temps w/RealTemp when I set the system up.
 
cool, I need this stuff , I am going to overclock.....

i7- 920


can I just google and download these programs, correct?


thanks!

al
 
uh oh..I get yelled at in other forums by using tooo much spacing...bad habit!...thanks for the info! big al
 
burn test gets it the hottest but passed a 20 session burn test that put my system temps in the 75C area and prime crashed my system in 30 minutes. Tested it out with game thought it was just prime and sure enough I was unstable.

So i use prime for over all stability and use burn test just to make sure temps don't go above 80.
 
I prefer Intel Burn Test because I usually don't have the time/patience to test my system for 24-48 hours with Prime95. IBT is a good enough program to usually determine stability in a much quicker time span.
 
I'm all for the Intel Burn Test too. There is a new version out with GUI interface. Anthing to get away from the 8, 12, 24hr ... testing time, what a pain in the neck!!
 
burntest.png


Ill assume this is a good sign :)
 
Technophobe said:
I'm all for the Intel Burn Test too. There is a new version out with GUI interface. Anthing to get away from the 8, 12, 24hr ... testing time, what a pain in the neck!!
Click to expand...

Thanks for letting me know. I'll check it out.
 
PC_User said:
Thanks for letting me know. I'll check it out.
Click to expand...
I linked to it in my post (#2), so you don't have to look for it.

criccio said:
Run it more then once. My system couldn't get through 5 in a row.
Click to expand...

I do a min of 20 passes (with 8GB of RAM), anything less than 20 I just don't feel good with, I have had it do fine all the way to 15 passes and freeze up on the 16th.
 
25runs.png


Here it is after 25 runs. Still looking good. I want to try to overclock more now but I dont know if its even worth it at this point.
 
Have you tried RealTemp? For some reason, I always got temps which were 10*C lower with CoreTemp...
 
PC_User said:
Have you tried RealTemp? For some reason, I always got temps which were 10*C lower with CoreTemp...
Click to expand...
That would only happen if your Tjmax was set incorrectly in Core Temp. The current versions of Core Temp and Real Temp use the same Tjmax values for Core 2 and Core i7 CPUs so they will both report identical temperatures.
 
Yup yup, those programs have to sort of guess what's the correct Tjmax value (because Intel doesn't publish them, but they're sort of agreed on by the community at this point), and they extrapolate temps from the value reported by the CPU compared to that... The CPU's sensor doesn't report an actual temp per se, it reports distance to Tjmax from what I understand. That's why load temps are the most accurate and idle temps can many times be considered irrelevant, the farther from Tjmax you go the less reliable the reading is.
 
CPU or CPU+Mem burns are good for a short term OC test but I consider a long gaming session the most accurate test.
 
Not really... A long gaming session won't stress all cores simultaneously and a long gaming session might not even bring your GPU temps as high as some of the synthetic GPU stress tests can (unless you can stomach an 8-hour Crysis binge).
 
I run both Prime95 and IBT to test for stability. I've used OCCT in the past, but I like the other two.
 
i use prime95 and for my video card i use some program actually called video card stability test.
 
Head over to the Distributed Computing section and see how to set up 2 SMP clients in a Virtual Machine - That will test the full stability :D
 
Impulse said:
Yup yup, those programs have to sort of guess what's the correct Tjmax value (because Intel doesn't publish them
Click to expand...
Actually, Intel did publish them. The current Tjmax values for Intel CPUs are known. That's why all the newer versions of the temp-monitoring programs report the same temperatures.
gwarren007 said:
Head over to the Distributed Computing section and see how to set up 2 SMP clients in a Virtual Machine - That will test the full stability :D
Click to expand...
Or rather two SMP clients plus a GPU client (what I'm running now) :p. Pegs all four cores at 100% constantly.
 
Ahh, that's good to know Zero82z... I remember all the initial speculation based around the mobile processor Tjmax values (which were the only ones known at the time IIRC), there's probably still a hundred threads floating around on the subject. I didn't realize Intel had finally published 'em, that's re-assuring. Speedfan's probably using the same values too huh.
 
Yea, speedfan, coretemp, and realtemp all show 100c as the Tjmax for my Q6600. Didn't always used to be that way with the older versions.
 
Zero82z said:
That would only happen if your Tjmax was set incorrectly in Core Temp. The current versions of Core Temp and Real Temp use the same Tjmax values for Core 2 and Core i7 CPUs so they will both report identical temperatures.
Click to expand...

Must have been using an older version of CoreTemp.
 
Impulse said:
Not really... A long gaming session won't stress all cores simultaneously and a long gaming session might not even bring your GPU temps as high as some of the synthetic GPU stress tests can (unless you can stomach an 8-hour Crysis binge).
Click to expand...

And a cpu/mem burn test won't test your graphics/bus bridge, your northbridge or your southbridge. You aren't testing stability, then.

You do make a good point that a game will probably only load one core but you can load additional applications, or multiple sessions of the same game, on to the other cores.
 
As long as the system handles everything you throw at it without any crashes, do you really need it to be 100% load stable? How often do we load up our CPU/RAM 100%?
 
Actually, I have a different problem. My computer handles all the stress tests with no problems, but will randomly stop displaying the SATA drives...go figure.
 
NoEcho said:
And a cpu/mem burn test won't test your graphics/bus bridge, your northbridge or your southbridge. You aren't testing stability, then.

You do make a good point that a game will probably only load one core but you can load additional applications, or multiple sessions of the same game, on to the other cores.
Click to expand...

There are stress/burn tests for your graphic card too, which you can run alongside Prime95... CCCT includes several tests, I think one is a version of Furmark, which is another good one, and there's always looped 3dmark runs if nothing else. You can run any of that alongside Prime95 which will just pick up spare cycles across all cores, so you load everything 100% and you get an accurate picture of the worst case scenario for your system, both temperature and load-wise.
 
Kendrak said:
Nothing pushes a system harder than folding.
Click to expand...

Based on what? You mean if it's running in the gpu as well as the cpu or you just mean in general? Because a) that's not true, in either case [memory isn't stressed, the hard drive is rarely accessed and graphics aren't touched unless you're running the gpu instance], but b) if you're saying that distributed computing, of which folding is one example, is a good way to test system stability then that's true... because it's at least a worthwhile use of crunch.

Even if you run a video stress prog I don't see how that's going to match the wholistic stress/useage of a game.
 
You must log in or register to reply here.
Back
Top