Sandy Bridge 4.5GHz OC

[Danzo]

Hey folks I've just pushed my sandy bridge 2700K from 4.3 to 4.5GHz air cooled by a Arctic Cooling extreme and I've got a Vcore of 1.262v.

What I would like to know is could i push this Vcore down even more?

Also I've circled the temps in a few programs i use during a stress test and I'm rather confused as to why I'm getting 3 different core temp readings. I know for a fact that the fan xpert is wrong but is real temp right? or is it Intel tuning utility?

If the core temps of 60ish are right, it looks like I've got plenty of head room for even more GHz do you think i could push 4.7 or more? Thanks

I tested the OC using Prime 95 (half an hour to an hour)
Intel extreme tuning utility (15-30 mins)
Real temp built in 10min cool down prime test
Hyper PI 5M decimal places
Cinebench CPU test

 

[Drangueos]

you can push more...I would keep the voltages under 1.3v... i know it is safe even more.. but i prefer not to push over 1.3v

but I think 4.5ghz is a good speed and 200mhz more will not make any difference, but will increase the temperature a lot.

 

[unclewebb]

Asus Fan Xpert 2 reports temperatures a long ways away from where the action is. It's useless information but I am sure it results in less phone calls to the help desk.

Intel XTU and RealTemp report data from the same core temperature sensors. It looks like you might have adjusted TJ Max in RealTemp. There is no need to do that with a Core i processor. Open up the Options window and click on the Defaults button. Most 2500K and 2600K CPUs use a value of 98C. It's possible that Intel dropped this down to 95C for the 2700K series.

Clicking on the Defaults button will read the correct value from your CPU. After you do that, it should be comparable to Intel XTU. The Intel software continuously gets data from the 4 sensors and then reports the highest value as far as I know.

RealTemp T|I Edition
http://www.overclock.net/t/1330144/realtemp-t-i-edition

 

[Threepwood]

I would run IntelBurnTest on High settings, then start up furmark on maximum settings at the same time.

You cant use maximum settings on IBT because there is not enough processor power left to run the GPU at maximum load.

This will give you a better idea of maximum system heat generation, along side power draw/stability when your PSU is powering other things.

 

[Doomedspeed]

Sandy has a high voltage limit than Ivy (which is 1.3)... Sandy's limit is 1.38-1.4, i think they revised from 1.38 to 1.4 just before ivy came out.

So you have a lot more room to move.

What i'd stress with is.

Prime (30 minutes - 1 hours) - You do this anyway .
IBT (30 minutes) - You want temps to stay below 75C (IBT gets chips hotter than any other program).
Game (30 minutes- 1 hour) - Make sure it doesn't crash..

Keep Vcore below 1.4 and you'll be fine.. From the sounds of it so far it could be a 5Ghz chip.

 

[robj20]

Your lucky, my 2600k doesnt like anything over 4.3Ghz which requires 1.296v. 4.4Ghz crashes even with 1.32v.

 

[BS Dave]

my 2700k needs 1.4 for 4.6Ghz - which is what it's getting. though that's with LLC at medium - ie not very much. if i bump it high i'm stable at 4.7.

 

[wazza300]

@ the op you need to use a more stronger stress program imo,and use realtemp or coretemp to monitor true cpu core readings

I like to do 5 runs in ibt max stress test to see where I am with temp limits,ideally under 80c

for long term stability just use pc as normal or gaming and tweak voltages if any bsod's crop up

 

[Danzo]

you can push more...I would keep the voltages under 1.3v... i know it is safe even more.. but i prefer not to push over 1.3v

but I think 4.5ghz is a good speed and 200mhz more will not make any difference, but will increase the temperature a lot.

Yeah guess 4.7 isnt much difference but at least I can brag to my mates who think their awesome because they have a Pilledriver with a stock clock of 4.4 hah

 

[Danzo]

Asus Fan Xpert 2 reports temperatures a long ways away from where the action is. It's useless information but I am sure it results in less phone calls to the help desk.

So is there anyway I can tell fan Xpert to read the core temps? because the fan profiles are reading Xperts temps which seem to be way off

Prime (30 minutes - 1 hours) - You do this anyway .
IBT (30 minutes) - You want temps to stay below 75C (IBT gets chips hotter than any other program).
Game (30 minutes- 1 hour) - Make sure it doesn't crash..

Keep Vcore below 1.4 and you'll be fine.. From the sounds of it so far it could be a 5Ghz chip.

Cheers I will give that a go IBT looks like It's quicker to tell me if I'm stable or not. It looks like i've got lucky in the CPU lottery because your right I may have a 5GHz chip

One other thing can I ask why my Cinebench scores vary so much? Sometimes I get 8.72 then other times I get 8.61 and sometimes even after clocking from 4.3 to 4.5 scores would be the same or lower? Then i get the occasional 8.72?

Cheers guys

 

[unclewebb]

Intel specifically located their core temperature sensors on the hottest spots on the core. That's the data that most programs report these days. This data is used to control thermal throttling and thermal shutdown of the CPU.

Peak core temperatures change instantaneously so trying to control a fan with that data might cause the fan to hunt up and down in speed. That might be the reason that Asus decided to use temperature data from a different sensor. There is no way that I know of to switch the Asus Fan Xpert 2 program to use core temperature data.

I found a difference of about 20C when my 3570K was fully loaded with Prime 95.

http://img138.imageshack.us/img138/2300/realtempaisuite.png

Intel CPUs do a great job of looking after themselves and can run reliably at most any temperature so no need to worry about the perfect fan speed or core temperature.

http://img27.imageshack.us/img27/6216/torturetest.png

 

[Danzo]

Intel specifically located their core temperature sensors on the hottest spots on the core. That's the data that most programs report these days. This data is used to control thermal throttling and thermal shutdown of the CPU.

Peak core temperatures change instantaneously so trying to control a fan with that data might cause the fan to hunt up and down in speed. That might be the reason that Asus decided to use temperature data from a different sensor. There is no way that I know of to switch the Asus Fan Xpert 2 program to use core temperature data.

I found a difference of about 20C when my 3570K was fully loaded with Prime 95.

Wow 101 C is pretty crazy hot for your i5? :O cheers though it was just because i was playing battlefield the other day and the CPU was going upto about 75C without any reaction from the fans when the profile was set to be 100% fan load at 75C but Xpert was only detecting 50C should i just off-set my fan profile to work with the lower reading of Xpert?

thanks

 

[unclewebb]

It's easy enough to do testing similar to what I did above so you can compare core temperature to what the Asus Fan Xpert software shows for your CPU so you can come up with a suitable offset value. 20C is usually a good ball park value depending on what heatsink and computer case you have.

The only question I have is why bother. The picture above shows that Intel Core i CPUs can run reliably at some extreme temperatures. Your peak core temperatures are well, well within the Intel spec for your CPU so why have your fans blazing trying to make it run even cooler? You already have enough head room that you can probably bump your CPU up another 200 or 300 MHz without even having to think too much about it. Just up the voltage as much as you need to so that your CPU remains Prime / IBT stable.

I had to add some extra voltage while using the Intel OEM cooler to get my 3570K up over 100C. I was also giving the on die HD4000 GPU a good work out too. I just like making sure that Intel builds good stuff and they do!