OCUK Sandy Bridge, SB-E and Ivy Bridge 5GHZ Club

Minstadave said:
What cooling is everyone using?
Click to expand...

I'm using prolimatechs rev B, temps never get past 80C at full load, with 2 fans i can lower that by 3-4 degrees but its a bit too noisy
20104774.jpg
 
Last edited:
SeeknDestroy said:
I'm using prolimatechs rev B, temps never get past 80C at full load, with 2 fans i can lower that by 3-4 degrees but its a bit too noisy
20104774.jpg
Click to expand...

What fans are you running?

I've got that cooler with 2 Akasa Vipers and my 4.5 @ 1.31 maxes out at 63c at the moment and i'll be aiming for lower this weekend.

I'm going to have another go at 5ghz I think but with some tighter ram timings. my SuperPi 1.5 time is pants compared to everyone else :p
 
Todge said:
What fans are you running?

I've got that cooler with 2 Akasa Vipers and my 4.5 @ 1.31 maxes out at 63c at the moment and i'll be aiming for lower this weekend.

I'm going to have another go at 5ghz I think but with some tighter ram timings. my SuperPi 1.5 time is pants compared to everyone else :p
Click to expand...

Scythe Slip Stream, they are very good but somewhat noisy at full speed. That's why I've decided to run only one of them
 
dave_beast said:
Using Cooler Master Hyper 212+ with push/pull, never get over 68ºc load since adjusting cpupll to 1.7v
Click to expand...

Thats not bad at all. In my case, my temps have a very wide range. For example when I'm outside and the heating is off, I leave the computer running prime and when I come after 4 hours I get peak temps of 72C.
However with the heating on they're always 78-80. As I said previously, I'm using only one fan though
 
EniGmA1981 said:
So in theory, if I can manage my temps, I could run my C2D at 2v at 5GHz?

In theory of course... lol
Click to expand...

:D:D No not even in theory. There are certain natural laws at play, one of the is the law of thermodynamics. And if you put that much energy into the silicon wafer of your CPU then the temperature will not be manageable. And anyhow you do not need that much voltage to run at 5GHz.

But you are close to the truth - for example when we use LN2 on our 980X CPU's we have them down to -180'C and people will put 1.9 volts into a CPU that costs £780 :D

Rule of thumb if you want 5GHz on air with this CPU you will need 1.440 volts. All the rest can be on auto.
 
1Day said:
Heat kills - not voltage. Voltage makes heat. So manage your heat.
Click to expand...

Sorry but I don't agree with that. Heat output is a result of frequency and voltage, not either or. You can't put 5v through a CPU if you can keep the temps sub-zero.

CPUs are designed to work at a certain voltage range, if you remove heat equation using something like phase change you can still over volt a CPU and kill it. So really both heat and voltage are key factors here, if anything I'd say voltage is far more important, with the right kit you can manage the heat output.

You also have to bare in mind the die shrink from 45nm to 32nm means everything is on a smaller scale and generally requires lower voltage to operate.
 
I think you missed my reply two posts up mate.

Let me quote what I said.

There are certain natural laws at play, one of the is the law of thermodynamics. And if you put that much energy into the silicon wafer of your CPU then the temperature will not be manageable.
Click to expand...


Increase in frequency does not produce heat. Not in itself. It is a common misnomer, but I suspect it is because most of the time when frequency is increased it is because voltage has been added to facilitate the increase in frequency.

But it is easy to test.

You take your CPU and set your voltages at a fixed amount.

Then take a reading of the temps.

Now increase your frequency of the CPU without changing the voltage at all.

Let us know what you find. I guess you will find your results match what I have found when we tested with students. No increase of temperature.

You see heat is energy and energy can not just magically appear it needs a source. Thermodynamics as I said earlier. So to have any increase in thermal product or heat it takes an corresponding amount of added energy, therefore there needs to have been a source that provided that additional energy.



But I do agree with you. :) If you push too much energy into a CPU yes you will kill it. And since my comment is causing some confusion I shall remove it if you prefer.
 
1Day said:
I think you missed my reply two posts up mate.

Let me quote what I said.

Increase in frequency does not produce heat. Not in itself. It is a common misnomer, but I suspect it is because most of the time when frequency is increased it is because voltage has been added to facilitate the increase in frequency.

But it is easy to test.

You take your CPU and set your voltages at a fixed amount.

Then take a reading of the temps.

Now increase your frequency of the CPU without changing the voltage at all.

Let us know what you find. I guess you will find your results match what I have found when we tested with students. No increase of temperature.

You see heat is energy and energy can not just magically appear it needs a source. Thermodynamics as I said earlier. So to have any increase in thermal product or heat it takes an corresponding amount of added energy, therefore there needs to have been a source that provided that additional energy.

But I do agree with you. :) If you push too much energy into a CPU yes you will kill it. And since my comment is causing some confusion I shall remove it if you prefer.
Click to expand...

Yes I was just commenting on your stand alone statement, I admit I never read your earlier comment. Which others may do so I was just clarifying things. You obviously know a lot more about the science, but it pays not to make sweeping comments which may confuse laymans like myself! ;) :p
 
Speed consider it done. The statement which we are discussing was made within a context of chatting about coolers - but as a stand alone it is fraught.


**edited my previous post.
 



this was with 1.4V on a P8P67 board. The system runs prime 95's short FFTs for 15mins without issues but I get BSOD after 30 secs of the blend test. All this with the memory set at stock speed.
Has anybody got both of these tests running stable with a voltage close to 1.4V? I certainly don't wanna increase it much higher until we get some more feedback on the actual max safe voltages for the sandybridge.
 
Last edited:
If you pass fft but fail blend, then it's ram / mem controller related.
I'm mobile at the moment, but take a look at my memory thread in the CPU section.

I don't really understand it but it seems to be some of the backend settings that make the difference. I can set my ram at 1866 and it will memtest and fft but not blend.

With mem at 1600 I'm stable in both blend and fft right up to 5.2
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom