Mrboo And His Stupid "mild" Overclocking Troubles Again

[markiemrboo]

My stuff has arrived and it's all up and running. I'll cut straight to the chase.

 

I can do 400MHz x 7.0 = 2.8GHz, stock vcore (1.225v, but CPU-Z reports 1.15v under load) and it ran Prime95 for 12 hours without fail. RAM is stock at 400MHz 5-4-4-15 at 2.0v as per SPD.

 

I tried to get a nice round 3.0GHz, which was then just a case of adding +0.5 to the multiplier. Prime failed on one of the cores pretty quickly. So I increased the vcore to 1.250v. It lasted about a minute, then spontaneously rebooted! Since I didn't see any fail messages, I tried increasing the "FSB" voltage by +0.1v (the minimum). It seemed to last a tad longer, about 15 minutes maybe, but then spontaneously rebooted on me again. So I put that back to stock and tried MCH voltage by +0.1v. Same. Tried PCI-E by 0.1v. Same. Tried PCI-E, MCH and FSB by 0.1v. Same.

 

I put these back to normal, and increased the vcore a tad more to 1.2775v. CPU-Z shows this as 1.200v under load. It managed about 3 hours and then rebooted during an online game of scrabble with my mother Great. So at 11:30pm last night I increased to 1.285(?)v and started Prime. This works out as 1.218v under load according to CPU-Z. I woke up and it was still going at 7:30am when I left for work. I came back to find it sitting at the logon screen. Great. The log showed the last event at about 8am, and no fails. It just rebooted again it seems.

 

Temperatures at these settings, from RealTemp 2.5, logged 66C max. Not a clue how accurate that is, but it seems kinda high. The heatsink doesn't even feel warm, and it bends the damn motherboard so I am fairly sure it's well seated! It was a royal pain to fit, but all 4 pins made an audible click eventually. I'm not sure how comfortable I feel increasing vcore even more for an extra 200MHz, especially with those temperatures.

 

So, before I give up and settle at 2.8GHz, which, don't get me wrong, I would be very happy with, but a nice round 3GHz would be very very very happy indeed, does anyone have any ideas how to get it stable?

 

- Temperatures are quite high aren't they? I've read reports of 45nm chips having dodgy temperature readings, which may just be people not able to accept it's so high and playing the blame game?

 

- Is CPU-Z vcore generally quite accurate? If so it's below stock voltage under load and at 1.285v?! That seems.. fairly bad. I've read conflicting reports, some say it is accurate, some say it isn't.

 

- The memory is at stock at these settings, as far as I am aware. I only changed the CPU multi and vcore. Memory is still 400MHz, 5-4-4-15 @ 2.0v and passes some quick memtest86+ testing.

 

Oh, I should add that it's not blue screening. I always turn off the automatic blue screen reboot option. It's actually really just spontaneously exploding

[CanadaGUY]

The value you set in bios for Vcore, will almost always be alittle higher then that which is showing up in CPU-Z. It's called Vdroop, there are mods for mobo's to help with how big a drop it is, but it most likely will always happen. Why, well ask the electrical engineers that question

 

You might want to grab another cooler, are you using an aftermarket one or just the stock? 66C seems pretty high for the amount of OC/Volts your giving that cpu. Other then heat, you should be good to up the voltage a tad more (most say like 1.5v safe, don't quote me on that, just what I've picked up from being around), as long as your not going any higher then 66C load. *Even that makes me kinda wary, I try to keep my load under 50 at all times.*

[markiemrboo]

The value you set in bios for Vcore, will almost always be alittle higher then that which is showing up in CPU-Z. It's called Vdroop, there are mods for mobo's to help with how big a drop it is, but it most likely will always happen. Why, well ask the electrical engineers that question

 

You might want to grab another cooler, are you using an aftermarket one or just the stock? 66C seems pretty high for the amount of OC/Volts your giving that cpu. Other then heat, you should be good to up the voltage a tad more (most say like 1.5v safe, don't quote me on that, just what I've picked up from being around), as long as your not going any higher then 66C load. *Even that makes me kinda wary, I try to keep my load under 50 at all times.*

 

From sig: Intel C2Q9300 + Scythe Zipang

 

I have read 1.35 - 1.4v should be safe on a 45nm chip... but with these temperatures? I'm thinking no... :-S

[kingdingeling]

Maybe the chip doesn't like the half multiplier? Have you tried 7x429?

[markiemrboo]

Maybe the chip doesn't like the half multiplier? Have you tried 7x429?

 

I hope it does, because 7.5x is stock But in all seriousness, no, I haven't tried that. I need a quick way to make this magic reboot happen really :-D

[road-runner]

I am going to say the OCZ RAM is stopping you. I have trouble with my OCZ RAM much above 400 it gets flakey. You are running the voltage recommended for the RAM?

[markiemrboo]

I am going to say the OCZ RAM is stopping you. I have trouble with my OCZ RAM much above 400 it gets flakey. You are running the voltage recommended for the RAM?

 

It's at stock as far as I am aware. 400MHz, 5-4-4-15 and 2.0v. That's what the EPP SPD profile says in CPU-Z, and those are also my current settings.

[road-runner]

I know just to get 3.9Ghz out of my QX9650 I have to set it on 1.45 in the bios, it droops to 1.392-1.4 under load....

[markiemrboo]

Maybe I just need more vcore / have a bit of a rubbish chip, or high expectations

 

I've set it to 1.300v. CPU-Z reads 1.232v under load. Increasing vcore is the only thing that really seems to have helped so far. I'll have to decide whether the extra 100mV and 6C of heat is worth the extra 200MHz ( assuming it's now stable, and i'll bet you anything it wont be )

 

66C max is pretty poo I have noticed that one core is always about 5C lower than the rest. The brilliant thing is.... it seems to change core! One day it's core #2, the next it's core #4. Strange things.

[road-runner]

The quads require a good cooler and good mounting pressure. I have tried the pushpin coolers and they dont cut it. I was getting around 64-67c with a typhoon and the stock fan at stock volts and 3.2Ghz. I took the push pins out mounted it solid with a backing plate and put a 88cfm yate loon on it and its running around 60c full load at 3.6Ghz and the temps are pretty even now. When it was loosely mounted with push pins temps were real uneven and high. Pushpins dont have enough mounting pressure from what I have seen....

[markiemrboo]

Tosh This thing bends the motherboard nicely with push pins. It was scary as hell to fit...

 

[markiemrboo]

It may possibly be rubbishy case airflow with regards to high temps

 

I have one 120mm in the front, one 120mm in the lower side (right next to the graphics card) and one 120mm in the back. I had to remove the 120mm from the back, as the heatsink is now in the way, so I am down to just two case fans for the moment. All fans are at 7v to try and keep things quieter too. I've just fitted an old 80mm fan at 7v in to the upper side panel bit, and temps now seem to be hovering at 58C.

 

I'm trying to find a slim 120mm to stick in the back. 120mm x 20mm should fit. Just. The only one I can find in the UK, a Yate Loon at 20mm thick, is out of stock If anyone can help there I would appreciate it.

Edited May 6, 2008 by markiemrboo