View Full Version : Article: Overclocking Intel CPU's
A very interesting read, anyone with a recent 45nm Core2 CPU needs to read this before they O.C.
http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3184&p=1
Excellent find Cleggy. It's articles like these that show that Anandtech is in a league of their own. Tom's does have some talented staff but a couple of absurd results and conclusions I've heard from them discourages me from making them my main source of reading.
I'd like to reiterate this
We've also learned how rather than pushing a CPU to the ragged edge, sometimes it is better to find that point where performance and efficiency can co-exist in a delicate balance. A good overclock is not always about maximum speed. Instead, we would like to advocate what we call the "intelligent overclock" - find that point where going higher just doesn't make any sense…then simply stop and tune from there. We have provided many of the tools for doing so in this article; now it's up to you to make it happen.
As it's all too important.
Its stuff like this that amazes me, so much testing must have been involved...
During the course of our testing we made a rather interesting discovery regarding 45nm silicon scaling: a window exists in which CPU frequency responds in a highly proportional manner. Calculating this value later tells us that between 3.0GHz and 4.0GHz our processor requires ~0.3mV (0.0003V) more Vcore for each one megahertz increase in core frequency. Since our QX9650 is capable of running the stock 3.0GHz setting at only 0.98V, this means that achieving a stable 3.6GHz overclock requires 0.98V + (0.3mV/MHz)(600MHz) = 1.16V. This general trend continues all the way to about 4.0GHz where we found total stability at an amazingly low 1.28V. We cannot help but feel excited about Intel's new 45nm process, especially considering such early maturity.
http://images.anandtech.com/reviews/cpu/intel/penryn-oc2/Power_Equation.png
Also useful :)
We can calculate the absolute CAS (Column Address Strobe) delay in a few quick steps. DDR-800, which is in fact double date rate as the name suggests, runs at a base frequency of 400MHz or 400 million cycles per second. Inverting this value tells us the number of seconds per cycle (2.50ns). Finally, multiplying this by the CAS rating tells us the total delay time of 7.5ns (3 x 2.5ns). Likewise, setting a CAS value of 4 results in an absolute CAS delay of 10ns. We can see now why higher CAS values give way to lower memory bandwidths - in the case described above the MCH spends more time "waiting" for data to become available when the memory is set to CAS 4.
wonderlust
13-04-08, 14:17
Makes me glad I stuck decided to stick my q6600 to 3Ghz for most of the time, no voltage increase required
i generally tend to stop when it requires a big bump in voltage to get it stable at the next "step up in speed" if you know what i mean?
wonderlust
13-04-08, 14:41
yep, no point in cooking a chip to death, mine needs a good 1.5V to do 3.6Ghz and I think that's just too much
i generally tend to stop when it requires a big bump in voltage to get it stable at the next "step up in speed" if you know what i mean?
Yeah it's the sensible approach really. My Athlon 64 San Diego's could make it to 3.0-3.2GHz with plenty of volts but always stuck to 2.8GHz as it could be done at stock volts or even an undervolt. Same with my Opteron 165 and also my Core 2 Duo E6300 which does 2.8GHz with a 0.15V undervolt.
wonderlust
13-04-08, 15:24
the one downside of the Msi P35, you cannot undervolt :(
the one downside of the Msi P35, you cannot undervolt :(
plz explain?
it sets it to vid and that as low as it goes?
That seems to be what he means yeah. i.e.- if VID is 1.25V that's his lowest option.
wonderlust
13-04-08, 16:20
Yep, The lowest Voltage I can select on my Q6600 is 1.3125V
If I go to the next setting down it loops to 2.1v which is the maximum my board would provide to this chip, (hmm wonder how long it would last...)
thats not too bad, its on less than that now and its solid @3.2 so....
Ive just had an arguement with one of my colleagues at work and he says overclocking is stupid.
Basicly what he says is intel test all there cpu's at certain speeds and find out what they safely run at, and therefore overclocking them greatly reduces there lifespan. No matter how cool its running.
He says hes studied electrical theory
Anyone wanna... well i dunno. what do you think?
wonderlust
18-08-08, 17:26
Yes he is not wrong, BUT Inte warrants a retail CPU for 3 years electomigration (i believe thats what its called) will have an effect but it will probably reduce a chips working life but to be honest it may reduce a 10 year life to 5 years, how many people still running 5 year old cpus?
I had a Celeron 300a when they came out, 1998, it has run at 50% overclock since the day it arrived (450mhz) and although I don't own it any more it is still running very happliy in the same board I built it in, just owned by someone else's kids
He's right, generally speaking it does reduce the chips lifespan.
But it will still go for a couple of years and by then anyone who is enthusiastic enough to overclock in the firstplace will have already upgraded.
But what hes saying is you may get a chip that could die in about a week...
wonderlust
18-08-08, 17:35
I would imagine Intels QC is better than that. If a chip was that bad it would fail very soon at stock.
They way I guess Intels QC works is that the test it as a q6700 and it fails then they test it at q6600 (obviously they start higher and work down).
But if they have a glut of Q6700 and there is a demand for Q6600 they are not going to go and make new q6600 they will remark q6700s and sell them as q6600s.
In the Coppermine P3 days iirc yeilds were so good that they were selling P3-933s with as p3-700s just be marking it such and reducing the FSB from 133 to 100.
I used a couple of these 700s in my server @ 933 for over 4 years :D (until I sold it for an AMD x2 :D
Yeah, basically what everyone is saying is right.. It will reduce the life of the chip, but not by any amount that will make it noticable - by the time you reach the end of its overclocked life, technology will have moved on and it wont be capable of running that much anyway (if things keep going as they are)..
And yes, you could kill a chip in a week if you're not careful and make sure you read up on what you are doing. But then by the same rule, you could also kill a chip in 3 seconds if you shove 2.5v through it and switch it on. As long as you're careful and know what you're doing, its fine. And if you dont know, then thats what these type pf forums are for - to learn :)
And what wonderlust said in the previous post is exactly right too :)
Thats what he was saying, he was adimant that it was a stupid idea putting more volts through a cpu, and he was saying that it could kill it very quickly.
I went on to explain that i know it can kill a chip, thats why you dont do stupid overclocks.
But i was wrong obviously as he was a fat ****er... (EIDT... is a fat ****er)
T also went onto explain that you only up the volts in tiny ammounts and most processors dont go to good with more than 1.4??? (if i remember correctly, or as you people like to put it iirc :P)
Off the top of my head, the stock voltage is 1.25v (someone correct me if i'm wrong) - but Intel states that they are capable of handling up to 1.4v (AFAIK). I'm not even near that voltage to get 4Ghz. I personally would be happy going up to 1.45v but wouldnt go any more than that. So yes, he is right and wrong at the same time..
At the end of the day, its his choice. Buy a 3Ghz chip and run it at 4Ghz, but maybe reduce the life a small amount. OR, buy a 4Ghz chip, dont reduce the life of it (but it'll be out of date by the time it gets to the point it would matter), and pay a fortune for it..
I know what i'd choose/chose... ;)
Moot point for most on here i would guess, personally i rarely keep a cpu for more than 18months tops so i run it a fast as i can with sensible settings.
Useable lifespan for most pc hardware is pobably 5 years absolute max, actual life span is probably more than double that, however with overclocking you might actually extend its "useable" life span.
Hey people, finally gonna have a shot at OC'ing my q6600
Any recommendations on what i should change?
I had one go so far, changed the FSB from 266 to 280! **** yeah a whole 100mhz increase
Not sure what else i will have to do, will the CPU cool itself automatically or do i need to install software to manage it?
Any ideas very much appreciated as always people
wonderlust
17-12-08, 09:08
try increasing it one a bit at a time, remember you may need to change your memory multiplier too.
Mine is very happy @ 333x9 for 3Ghz at default voltage but it will go to 3.6Ghz if i bump up the voltage. YMMV
ahh ok... whats YMMV?
And i was just gonna take it to 3Ghz
Seems to boot alot quicker already on just 2.7
http://i319.photobucket.com/albums/mm449/DDevilMonkey_666/Cosmos/CPUz.gif
Is it normal for the FSB on this screen to alternate between 333 and 334?
wonderlust
17-12-08, 11:45
yes there is often a minor difference in the clokc gen chip.
ymmv = your mileage may vary
Powered by vBulletin® Version 4.1.12 Copyright © 2012 vBulletin Solutions, Inc. All rights reserved.