The 5GHz race - Crosshair Z / 8320BE

[lismo]

Hi.

Last sunday i spend half day optimizing & tweaking my 8320BE up to 4.8Ghz. At first, one of those clips that hold the waterpump for my dual-rad AOI cooling snapped off. The thread was all worn, so i dediced to replace all 4 thumbscrews with real screws that stick out 1 cm out of the backplate. It has a good result, because i archieve 49 degrees on IBT standard and 51 degrees on IBT high. Scores are low yes, as i'm not done yet with tweaking. However, i'm currently stuck at 4.8Ghz on 1.475V which runs perfectly fine:



But 5GHz is a complete different quest. I need up to 1.56V at least in order to get it stable on standard. And sometimes the system completely freezes up upon stressing.

So i'm not sure what causes this. At first i followed the guide on https://rog.asus.com/articles/hands-...r-v-formula-z/ which is pretty easy, put some settings on, and start tweaking from there. Its just very weird i need a voltage bump of at least 0.7 ~ 0.8V to get 200Mhz more.

I'm sure that any of the crosshair Z settings such as VRM might help, at the cost of more heat / more power being used. I have'nt done anything to that yet, but i will give it a try later this day. Any tips?

4.8Ghz on a 8320BE is pretty good at 1.475. Temps are within range (and proberly lots lower when gaming / working compared to IBT). I have'nt done any memory tweaking as well, and hell no optimized the HTT / CPU/NB speeds and all.

[BeepBeep2]

What temperatures are you at when stress testing at 5 GHz?
Your chip may just be hitting a wall, it might not be capable of 5 GHz stable. Not all of them are, my 8370E only does 4.7 at reasonable voltage and I had a couple FX-8150s that couldn't do more than 4.6.

0.7-0.8v increase would be a lot, don't know if the chip would live on that under LN2

Also, try Prime95 blend test - IBT seems easier to pass iirc.

[lismo]

I figured it out,

When you change CPU voltage on Crosshair Z, it leaves CPU-NB voltage on AUTO. When that happens it boots up to 1.4V for CPU/NB for some reason, which i manually set back to 1.175v. I figured out the less volts go inside the cpu the less heat is being put out.

I upped this a notch and the problems seem to be gone now.

Now it's stresstesting 5GHz fully to find it's best possible voltage.