Where can i find info on OCCT Error Codes

[LOUISSSSS]

title says it all

 

[LOUISSSSS]

From the creator of OCCT:
1- Thread stopped because another thread has stopped for some reason (mostly an error 4)
2- User interrupt
3- Not enough memory to run the test (should never happens as the GUI is taking care of that)
4- Calculation error detected

 

[LOUISSSSS]

i've recently switched my main stability testing program to OCCT (from P95)

i used to give P95 12-24 hour runs on Blend and 12-24 hour runs on Small FFTs. Blend tested my ram (supposedly) and Small FFT's tested my CPU (supposedly)

now, i'm using OCCT v2.0.0b.b01 (v2.0 beta 1) and i've detected errors within 1 hour that would take p95 over 6 hours to detect.
90% of my error detections have been within 10 minutes in OCCT.

Everyone should give this stability testing program a try.

Other programs i use:
-Memtest86+ v2.01 - Mainly used to test Ram/MB DIMM slots, not really for stability. mainly for RMA reasons.
-HCI Memtest - hasslesome to use, but supposedly detects memory errors accurately and relatively quickly. It is a single threaded program that can only stress 1gb of ram at a time so another instance must be open to test 2 or more gbs of ram.
-p95 Blend/Small FFT's - after i pass everything else i've mentioned above i'll run each for 12 hours or so overnight.

keep in mind everyone shouldn't be using only one single program to test for stability.

 

[runawayprisoner]

Agreed. And... I think you can just disregard the error messages. Just concentrate on the voltage graphs. If vCore is showing distortions, try tweaking PLL and FSB (VTT). If 3.3V, 5V or 12V shows something, try tweaking vNB. If all else fails, and all of the graphs show from very small distortions to none, try upping vCore a bit then. That's how I got this:

http://i235.photobucket.com/al...am/PC/OCCT39stable.jpg

 

[LOUISSSSS]

Originally posted by: runawayprisoner
Agreed. And... I think you can just disregard the error messages. Just concentrate on the voltage graphs. If vCore is showing distortions, try tweaking PLL and FSB (VTT). If 3.3V, 5V or 12V shows something, try tweaking vNB. If all else fails, and all of the graphs show from very small distortions to none, try upping vCore a bit then. That's how I got this:

http://i235.photobucket.com/al...am/PC/OCCT39stable.jpg

cool, didn't know that. do u think u can tell me more about the bolded section? or can u provide a link telling me more about it?

 

[runawayprisoner]

Mmm... it has something to do with GTL ratios. On some motherboards, you can directly modify these ratios. On some others, the only way to find the golden settings... is to try and keep on trying. And sometimes, higher volts are just overkill... and they may even cause stability problems. The explanation as to what those settings do belongs in another writing, I believe. What I'd like to tell you right now is to try... and try... and try. Not two motherboards will use the same settings, by the way. The golden settings for my motherboard to do the above are:

vCore: 1.3500v (it's real after Vdroop mod)
vCore reference: 0.59x (59%)
vPLL: 1.66v
vFSB (VTT: 1.46v
vNB: 1.67v
vNB reference: 0.61% (61%)

For 3.80GHz (475 x 8), it's even lower:

vCore: 1.3125 (real after Vdroop mod)
vCore reference: 0.63x (63%)
vPLL: 1.58v
vFSB: 1.32v
vNB: 1.55v
vNB reference: 0.63x (63%)

Why such high voltages are required for such clocks... is beyond me. I only know that I tried and tried... and finally got what I want. Or actually, I exceeded it a bit.