Interesting find

[supastar1568]

All,

Recently have been working to get my e8400 3.80 Ghz stable, truly stable. With my settings I have passed 100, yes 100 passes of Linpack, 24 hours of Prime95 small FFT, 12 hours blend, and 7 hours of memtest. All games play fine. I decided to experiement with the last stability utility on my computer, OCCT. I did an 8 hour "mix" test and got a Code 4, calculation error after an hour and 20 minutes. I find this to be very interesting. Any hints to what it may be?

I am thinking I need to up my northbridge. My CPU seems stable with the 24 hours small FFT and the punishing linpack. Memtest justs tests the memory itself. The "mix" in OCCT may require an uppage in northbridge voltage. What do you all think?

 

[TheUnk]

How much time had passed since you did your initial tests and then OCCT?

 

[nerp]

It's not too crazy for this to happen. Try upping the NB voltage and see if OCCT passes aftewards. Your reasoning sounds quite reasonable.

There are people who have said that they can pass 100 passess of linpack but fail prime after a couple hours. I personally don't consider something stable until it can pass every test for however long you run it. This often means giving up on your max desired overclock if you're not willing to keep upping voltages.

 

[supastar1568]

Originally posted by: nerp
It's not too crazy for this to happen. Try upping the NB voltage and see if OCCT passes aftewards. Your reasoning sounds quite reasonable.

There are people who have said that they can pass 100 passess of linpack but fail prime after a couple hours. I personally don't consider something stable until it can pass every test for however long you run it. This often means giving up on your max desired overclock if you're not willing to keep upping voltages.

I completely agree. I will try upping the northbridge.

TheUnk,

About 12-15 hours had passed since I ran the last tests. I restarted the computer to double check the bios settings before running OCCT. So it got a nice clean restart.

 

[supastar1568]

Yup, overclocking is definitely sketchy.

Increased the northbridge, OCCT passed. Same test on intel burn test... failed (which it had passed the same test before, except this time it has 0.05 more volts on the northbridge).

Think I may give up and crank the voltage to an even 2.0 for pissing me off.

 

[Dadofamunky]

Heck, I'd call it a day with the settings you initially posted about. Sounds pretty good to me. At some point, after a 9 hour OCCT test and a 12 hour Prime test, I got tired of stability testing; the system runs great and I buttoned it up. At some you gotta actually USE the thing.

 

[Drsignguy]

Well based on the information you ( op ) have given, it all sounds fairly reasonable. Only one thing, you can isolate your Ram from LINPAC and Occt. You can set your memory to like 256k if you wish in LINPAC. I would definitely try that first because, it does sound like the memory is causing the discrepancies. BTW, what is your ram speed with the OC you have obtained? By taking out the memory out of the Over clocking equation, it will help you obtain a stable OC with your Cpu and then later you can work on your Ram.

 

[supastar1568]

Originally posted by: Drsignguy
Well based on the information you ( op ) have given, it all sounds fairly reasonable. Only one thing, you can isolate your Ram from LINPAC and Occt. You can set your memory to like 256k if you wish in LINPAC. I would definitely try that first because, it does sound like the memory is causing the discrepancies. BTW, what is your ram speed with the OC you have obtained? By taking out the memory out of the Over clocking equation, it will help you obtain a stable OC with your Cpu and then later you can work on your Ram.

Good idea. I will try that, I have now set up a nice excel spreadsheet and am ready to go. My FSB is at 423, with memory multiplier of 2.50A, giving me speeds of 1057, I have PC8500 ram, which is rated at 1066, so it is within spec. The only other thing that may be a possibility, which I'm leaning towards, is my TRD setting. Default at this speed it is 12, I have it set to 7, which may be an issue. I will try to narrow this down to components first.


Anyways, I do all this since I am curious and will be good info for any and all of my future builds. I still use my free time to play games.

I just see it as I work, go to the gym, and go to bed. It gives me lots of time to run tests while I'm not at the computer.

I figure, hey I gotta go to work for 9 hours, why not run a stress test and get some new values.

 

[Drsignguy]

Quite the multi tasker eh?

Well now it seems you are on the right Track..stay with it and let us know how it goes.

 

[soccerballtux]

Wouldn't worry about it man. You are stable. I only test for 8 hours Orthos. You have found 3.8!

 

[supastar1568]

Originally posted by: soccerballtux
Wouldn't worry about it man. You are stable. I only test for 8 hours Orthos. You have found 3.8!

Yeah, possibly. I will experiment a little bit more while I am away from my computer. Really beating on this e8400