[Official] The Ivy Bridge Stable / Suicide Club **Guides, Voltages, Temps & BIOS Templates** Inc... - Page 96

Think about it for a second. If applications received the faulty computation after the cpu catches and corrects it (whea19), then what's the point of correcting it? Windows knows about the internal error because it listens for it (a programmer wrote code in the kernel to handle the error message broadcasted by the cpu.) Prime95 does not.

I need a favor. I can't get my 2600k to put out those errors at all, even though it is possible with SnB. Next time any of you get the error, can you try upping VCCSA and/or RAM volts instead of vcore to see if this clears it up?

what does my ram volts have to do w/ anything I don't have it oc'ed at all unless the XMP profile counts as ocing it? I think it just tightens up the timings 1 notch. and I haven't ever touched VCCSA more info needed please since you seem to be the smartest person on here usually.

Yeah, thats's how it used to work.  I remember Prime95 stopped working with icon turned into red color when it detected errors.  But this was back in Pentium4 / Athlon64 days.

well in my experience that has still happened a few times not recently or even on this chip I don't think but I have seen it say rounding error expected 4 but got 5 or something similar and it stops that worker and continues the other 3

I actually have a few questions, as I am pretty new to this. I am using my sig rig.

I am now at 4.3 GHz @ 1.22V with a max temp of 74-85-84-77 on a high IBT for 10 runs. I will be doing Prime95 after I make this post. These numbers don't seem to great, as I have seen many others getting 4.4 GHz with 1.22V with the same cooler.

I did try 1.25V at 4.4 GHz and the computer failed to start. I also get a BSOD at 4.4 GHz @ 1.215V.

I am not looking to absolutely max the OC capability of the chip, but do whatever is perfectly safe to do.

What should my next step be?
Should I increase the voltage or leave it?
Should I change another value?

I am also having some trouble with my RAM, which is in this thread:
http://www.overclock.net/t/1265910/samsung-2x4gb-asrock-z77-extreme4-help#post_17406158
That may or may not be part of the issue, if there is one. Screenshots of all of my settings may be found in that thread too (the CPU related ones might be moreso in the spoilers at the end). Note that my voltage and multiplier have changed to 1.22 and 43 respectively. I have also set the PLL Voltage to the minimum allowed of 1.586V and have set the CPU load line to 2

Thanks in advance,
Big Al

the voltage required for a certain frequency is not based on the cooler but the chip. so don't be too hung up on what others are getting as ivy is a huge lottery...
Edited by HardwareDecoder - 6/4/12 at 8:37pm

Maybe I should leave my 3770K @ stock.

Sounds good, I didn't know that. It makes me feel a lot better.

For Prime95, is blend the way to go? I am going to run it soon.

Also, what would you do in my case:
I could increase the voltage past 1.25V in hopes of getting to 4.4 GHz or more.
I was hitting 98 as a max while doing IBT on high, and 77 as a max when doing blend Prime95.


+rep for the help!

you dont want to run at over 90c for a daily OC I believe.... that is hot which cooler do you have? and Small FFT's should be better to test your CPU OC, while as blend is better to test the whole system, small FFT tests your CPU only, blend test cpu+ram

At my 4.5, 4.6, and 4.7 prime 12 hour stable/no whea errors settings, I can lower vcore by .005 to .01v (1-2 notches in bios) and run 28K or 40K ffts and reliably get whea errors every time within 30mins. Or run prime 15hrs, and get whea errors reliably, especially on smaller ffts.

Or another way of putting it, once I can run prime for 15+ hours without uncorrected cpu errors, ie prime rounding errors (+ may get some corrected/whea errors), then I still have to bump vcore 1-2 notches more, to run prime for 15hrs and not get whea errors, ie cpu corrected errors. And that is true for all 3 overclock settings. So if your prime stable, and getting whea errors during prime, you likely just need to bump up vcore 1-2 notches and then you will no longer have whea errors during prime. Maybe Ivy does a better job than other cpus of catching some cpu errors, creating whea errors, when on the brink of stability. I cant produce them on my wifes still 1366 build when running at unstable settings, and she has no record of them.

If your using a sensible LLC setting that still allows at least a little vdroop, and you are not using offset voltage for overclocking, then you should be fine once you can prime without whea errors.

If your using offset mode/power savings, you arent done after getting prime stable without whea errors. If you want to vary vcore/mhz on the fly using offset/power savings you have to test to makes sure vcore is sufficient at low loads as well. Otherwise, you can still get whea errors or even crashes if vcore isnt sufficient at varying or low loads, ie surfing/opening apps/gaming. I like Ikeths suggestion for testing lighter loads that he posted earlier.

But if you want to see a lot of whea errors, I had over 200 in a few hours benching 5.1ghz, guess which day looking at screenshot:



Edited by opt33 - 6/4/12 at 9:45pm