Best OC Stability Program / Verification Utility

[Spammeh]

I like to prime small fft / blend, occt and linpack to make sure it's stable as you can't seem to be sure otherwise

I love linpack's ability to save time but it doesn't seem to give reliable results on some machines according to posts on forums :S So it's really hard just to rely on one method.

 

[Tweakin]

You are right...I just never update or switched back to Prime.

I used to also use BOINC and figured if it would run, I was good. I believe differently now.

 

[geokilla]

Not to thread jack, but where can I find Linpack, and how long should I run it for? It sounds like it's more difficult on the CPU than Orthos and OCCT. Should I just stick with my Prime/Orthos then 95% F@H?

 

[Markfw]

Originally posted by: VirtualLarry

Originally posted by: Markfw900
And my new E7200 with ECS motherboard will not even run the SMP client for more than a few hours @ stock ! But it will run the GPU client forever (3 days at least so far). I atribute my problem to cheap hardware (the ECS board)

See, this is why friends don't let friends buy ECS motherboards!

It may not overclock, but F@H is another animal. I am runnign the GPU client now, and it was great @3166 for day, and I had to go down. I still say they are great VALUE boards.

 

[n7]

Originally posted by: geokilla
Not to thread jack, but where can I find Linpack, and how long should I run it for? It sounds like it's more difficult on the CPU than Orthos and OCCT. Should I just stick with my Prime/Orthos then 95% F@H?

http://www.xtremesystems.org/f...howthread.php?t=197835

That's LinPack packaged in a nice easy wrapper.

 

[GundamF91]

Definitely Linpack. If you are saying your system is stable even if it's not "Linpack stable" then you're kidding yourself. There's no "almost stable", there's only stable or not. If it wont' past Linpack 5 passes then the system is not stable. It doesn't matter if you pass Prime or Orthos or OCCT for multiple hours, if you fail one of the stability tests, then it's not stable.

Of course, most systems will work perfectly fine since they are never subjected to Linpack type of stress, so if you can live with almost stable, then there's no reason not to use the computer. But you just need to know that it's not fully stable when pushed to the edge.

For me, I've tested my system until it's absolutely stable under Linpack. Then I'd know I've pushed the system to the edge, and everything's OK. From that point, I'd dial back a few notches to reduce some unnecessary temp. Thus Linpack is the yard stick that system stability is measured with.

 

[Viper GTS]

I wouldn't consider 5 passes of linpack to be stable. I've been running with 100 passes using a full 8 GB of RAM. I've seen plenty of cases of failures midway through 100 cycles. Five is a joke, but if you can't even pass 5 then you are REALLY unstable.

Viper GTS

 

[Spammeh]

have you guys seen the posts from people who pass linpack but fail prime? it makes me kind of doubtful just to rely on linpack at the moment. Saying that I still make sure to pass 20.

 

[GrumpyMan]

Linpack for me, but I do use Prime, Orthos, etc. also. Never hurts to be sure. Passed all of them, my cores never got over 59 C. with Linpack.

 

[40sTheme]

I prefer OCCT. It's the one that causes hard locks/reboots when others don't.

 

[taltamir]

I found a new and unique way to stress out my intels...
OCCTv2 running in mixed mode (maxing ram and CPU at the same time @ 100%), then while it is doing that i will open a saved firefox64bit session of 10 windows, each containing 10 tabs (each is a web page), for 100 pages total.

Typically while OCing, I will get blue screens with OCCT if voltage is much too low, then if i raise it by a couple of notches i will start getting OCCT errors instead, first immidiately, later after some time. Then if i raise it a bunch of notches it will stop erroring.
At that point that machine appears stable and run stables on anything, but if i do the firefox 100 pages + OCCT on ram AND CPU at the same time (only two of the three don't work), it will blue screen for several more notches of voltage increase.

 

[TC91]

hey i got a question about linpack, could it be a stability indicator for ram as well as the cpu?

 

[VirtualLarry]

I can't run IntelBurnTest, it gives me this error:

Intel(R) LINPACK data

Current date/time: Mon Sep 01 23:36:04 2008

CPU frequency: 2.926 GHz
Number of CPUs: 2
Number of threads: 2
Parameters are set to:

Number of tests : 1
Number of equations to solve (problem size) : 17440
Leading dimension of array : 17440
Number of trials to run : 10
Data alignment value (in Kbytes) : 4

No runs is allowed for LinData\x86\linpack32.exe: not enough memory


Does anyone know how to get this program to run?

 

[nyker96]

orthos and occt are my choices, they seem to produce different kinds of errors. Sometimes one fails much sooner than the other. so I use both. but I try to do about 6-8hr of orthos at the end just to be sure.

Edit: a quick question for those of you using linpack, how long and how many tests do you guys recommend for stability? And is there a better GUI that tells you fail/pass like ortho? The standard IST seems to be just outputting numbers that must be matched visually to pass the test. It's not difficult but I would imaginge someone would have made it a bit more mouse driven like orthos. Thanks. Just going to try it out.

Edit2: just tried it for 20 min, seems to be pushing my undervolted 2160@2.7 to about 5C higher than orthos. Going to try 3Ghz later with a higher voltage.

Edit3: it's done with the 5 passes i gave it and actually just told me it passed. I think the text interface seems friendly enough.