What software are you using to test an overclock?

[Jumpem]

I don't know what is being used these days? Last time I set something up in 2006, I used CoreTemp, SuperPi, and Orthos.

 

[VirtualLarry]

CoreTemp is still useful, download the newest version.
CPU-Z, Prime95 25.9 (32-bit or 64-bit), OCCT 3.0.1 (Linpack, PSU test), LinX, IntelBurnTest.

LinX and IntelBurnTest are frontends to Intel's Linpack benchmark, which also functions as a torturetest program. OCCT's Linpack is built-in to the program (I think).

Orthos hasn't been updated, and since it's only good for dual-cores, it isn't used much anymore.

 

[Jumpem]

Originally posted by: VirtualLarry
CoreTemp is still useful, download the newest version.
CPU-Z, Prime95 25.9 (32-bit or 64-bit), OCCT 3.0.1 (Linpack, PSU test), LinX, IntelBurnTest.

LinX and IntelBurnTest are frontends to Intel's Linpack benchmark, which also functions as a torturetest program. OCCT's Linpack is built-in to the program (I think).

Orthos hasn't been updated, and since it's only good for dual-cores, it isn't used much anymore.

What are the difference sbetween OCCT, LinX, and Intel Burn Test? Do they all use this Linpack thing from Intel?

 

[Kenmitch]

Heres my testing strategy after I get her to post and load windows

Intel Burn Test....Very quick to error out and alot of heat generated. Great for finding the magic voltage quickly.

OCCT for the default 1hr run....Really good chance this one will pass if IBT passed but if not my findings are maybe another tick on the voltage and she'll pass. Once passed I change the options and test again which most likely will pass anyways.

Prime95 torture test for the long haul as long as you can stand it. I've never had prime fail yet once I got this far. Try to let it go for 24hrs or longer tho and if it's still going a very good chance its gonna be stable.

My final test is done after prime has been running for 24hrs or so. I let prime keep humming away and I then slam Intel Burn Test with the maximum memmory available setting on top of it. Then I'll start to open and shut windows, explore my hard drive, etc. Once IBT passes I'll start it again and expore my drives some more....Pretty much brings system to it's knees.

Hmmm....Maybe the last test is kinda exteme but it's kinda fun to see the system crawl to it's knees

 

[faxon]

OCCT has an IBT front end and it has it's own version of prime95 built in as well. the algorithm is basically the same, but it finds errors 10x as fast as vanilla P95 does, with a built in temperature readout that is as good as real temp as well. it also will output a graph for any of the data it can measure for the duration of your testing

 

[theAnimal]

Folding@Home 24/7

 

[Zenoth]

CPU-z to check voltages
Everest and SiSoft Sandra for memory-related benchmarks
Prime95, Orthos and MemTest for stability stress tests
3DMark 03, 06, Vantage for general benchmarks

And, finally, games, for the best real world "end result", since my PC is primarily used for gaming.

 

[MagickMan]

IBT to find the voltage sweet spot > 2hrs of OCCT > enjoy. That's stable enough for me.

 

[Shmee]

about right. I would consider 20 minutes of linx pretty stable, but apparently 10 minutes wasnt completely stable. This was on the Q6600 machine, I finally settled on 3.5 GHz @ 1.408 Vcore. Much easier than 3.6 GHz it seems, which got much too hot under stress with the required voltage of > 1.425.

 

[Markfw]

Originally posted by: theAnimal
Folding@Home 24/7

^^^^This is the best, and real.

 

[Shmee]

yup, at 1.4 V and 3.5 GHz, I lasted 10 min of linpack, but not 2 days of the SMP client. So I upped vcore 1 notch, and it is still folding now!

 

[imported_Scoop]

Since I'm on a dual core still, Orthos blend + ATITool simulataneously. No game stresses the system to this level so if this passes for a couple of hours, it's stable enough.

 

[Griswold]

Originally posted by: theAnimal
Folding@Home 24/7

You're wasting the projects time and resources by potentially delivering garbage. Make your machine stable and reliable then start doing scientific work. :roll:

If you cant agree, ask those who run projects such as folding@home what they think of overclocked machines in general and how much garbage results they get every day thanks to craptastic overclocks or people testing their overclock with the client.

 

[lopri]

Linpack is my new favorite.

 

[Munky]

OCCT, newest version is 3.1.0. Finds errors pretty quick.

 

[Spikesoldier]

two instances of orthos and one instance of hwmon

 

[Markfw]

Originally posted by: Griswold

Originally posted by: theAnimal
Folding@Home 24/7

You're wasting the projects time and resources by potentially delivering garbage. Make your machine stable and reliable then start doing scientific work. :roll:

If you cant agree, ask those who run projects such as folding@home what they think of overclocked machines in general and how much garbage results they get every day thanks to craptastic overclocks or people testing their overclock with the client.

Yes, I do stability tests FIRST, but sometimes all the tests say its stable. Then I run F@H, and 2 days later, a problem, so I back it down a little. NO it does not kill the project if you only to it once in the lifetime of your box, especially if you Fold 24/7 like I do.