The Official OC (OverClocking) Thread!

russ..... we already know that.

its called nehlam. and the x58 mobos.

 

We already have that more or less. All of the boards manufacturers consumer that support Intel have the same sockets.

I don't think that AMD is going to go for that.

You mean CIE and EIST or what we've been discussing. Evey board manufacturer supporting Intel has those same features. With AMD it's called cool and quiet.

 

well said but I believe that for the most part, most ppl in this thread that have their cpu OC'ed have at least ran OCCT for an hour and posted a screenshot to prove it. not everybody but most.

some don't consider that to be a true test for stability or a long enough test but it's better than reaching a certain OC and then just providing a screenshot of cpu-z.

 

Agreed! Although I don't think that an hour is enough or for that mater 3 hours. My thoughts are that a minimum of 8 hours of stress testing is required before one can assume moderate stability. I however generally exceed 10 hours without the test failing for stress testing.

 

Sophocles,

I think that's funny because that's exactly who's idea it was in the first place! If you remember back a while I mentioned overhearing part of a conversation between two guys that I assume worked for AMD about a 3 chip CPU that pretty much did just that, putting everything on one chip!

By same socket, I should have been more specific! I did mean each manufacturers, either AMD's or Intel's socket of choice for something like this, of course. I see far more usefulness and versatility in a configuration like this than I ever gave thought to, over 2.5 years ago. Still we know that AMD was working in it! They also have 3 core CPUs in production now! It's something to consider for the possible future.

To the Future and Onward,
Russ

 

i have seen other problems in the past.....
ran orthos for 12 hours...no errors
ran Mass Effect for 45mins and BSOD
sooooo in reality, you really have to use the computer after overclocking for it to be a truly stable overclock...
-im1992

 

That's why I don't just use orthos as a testing platform. It isn't realistic. You have to test all parts of a system - and you can only do that by surprise surprise, using it!

 

A degree of truth, but it depends on your tests. If someting a piece of your hardware to fail, then it wasn't tested. There is nothing that you can do in normal use of a computer that will match a decent stress test. My favorite is folding gromacs. Your creash means that you didn't completely stress test your CPU and memory. I will do up to two or three test and in at least one test I will set it to small ffts. I have never built a computer that has failed after stress successful passing a stress test and that's a lot of rigs. My wife still uses a Northwood 2.8GHz@3.2GHz that I built back in 2001. A friend of mine is borrowing an AMD Athlon XP2500 plus that I built at the same time.

Surprise surprise using a system for even serious gaming or encoding comes nowhere near matching a well run stress test. It is the difference between walking in the park and running a pentathlon.

 

I'm not saying you shouldn't run a proper stress test, but you should run BOTH. Orthos alone isn't sufficient.

 

@all
I am getting my q6700 this evening (if UPS is not being retarded). So I put a "window type" air conditioner and will make a duct that will connect to one side of the casing....then I shall overclock that q6700 to 3.6ghz (hopefully!).
I hope it works out well for me.
@Sophocles
Maybe my gpus overheated??? (still on stock cooling - no overclocking)
But it happened...and thats all I can say. I hope that the duct (read above) will solve any of my overheating problems!
-im1992

 

A 3.6Ghz quad will get HOT on an CNPS9700 - that's a lot of power running through a CPU.

 

well not if its a low vid!! hehe

 

I don't think you will have the heat problems that others seem to think you will. the zalman CNPS9700 is a very good cpu hsf and not that far below a TR U120E.

 

thanks!
i appreciate the kind words!
-im1992
btw, does anyone do folding@home here? wanna join my team?
As of now, i have about 60Ghz of CPU power dedicated to Folding@Home :-O
-thanks, im1992

 

Imran, your computer probably crashed because your GPU was the one that went unstable. Seriously, no game is going to stress the CPU more than Prime 95.

Get ATI Tool and run the artifact scanner to stress test your GPU.

 

i will do that...
thanks,
-im1992

 

This is very common, because a lot of these synthetic benchmarking apps test only a few key areas of our system. We can run a few apps to cover but may still leave a few things out. A big one so often overlooked and not often tested for its stability is the SB and other aspects of the storage sub systems such as RAID AHCI and or just plain old EIDE or ATAPI controllers as well as USB oand other system bus devices when put into the real world function. They have to be expected to work all at the same time as some given point, even if only for 3 secs.

aMD's "Cool And Quiet" is just their version of Intels "P" state control and this is why it is not so stable and less so on an AMD system since it is also NB chipset based and both BIOS and CPU controlled. After a certain OC level, the system can no longer keep up and remain stable. This is why ASUS and most other mobo makers now disable this (EIST) automatically as soon as you OC. ASUS used to allow OC to a certain FSB then cut off EIST.

C1E+TM is the few of these "C" or "P" state features that can be used i an OC'd system for at least the "C" states side of it. This reason is because as we OC a CPU beyond a certain FSB level, the "P" state values may not be good enough to keep stable.

On mobos that actually allow EIST to be enabled during an OC, or for stock users, use one or the other of "EIST" or C1E+TM2/TM1. They are competing to try to do a similar job and are only gonna' cause a slight lag in performance. For a OC'd system, it gets worse, it may freez and or crash during the struggle.

Oh ja, cool beens ons on the Q6700. If you duct the fresh cold A/C air in, be sure to get the exhausted case heated air out just as good. As long as you're using the window A/C, utilize its return intake to do this so the room doesnt take in all the heat of the system. Believe me, even my water cooled units heat up an average sized 12x12' office in a matter of an hour or so if I dont sent that exhausted air somewhere else.

As for your CNPS9700, I agree, you are still above average what most are using. It comes somewhat closer to the ThermalRight Ultra-90 of a single average fan.

 

True, but those areas include CPU, memory, chipsets, and FSB, and by their very nature all forms of PC operations are synthetic since they are all in the digital domain. Prime95 was never intended to be a stress testing application when it was first developed it is an application looking for Mersenne primes (really large primes). That is no more a synthetic application than gaming or DVD compression. The great thing about Prime95 is that it's really looking for data corruption, errors. Prime95 doesn't just stress a system looking for heat until it crashes it also stops when the smallest errors are reported.

If Prime95 ignored errors than a large number of failed systems would just roll on cranking out errors hour after hour. People with high overclocks who don't stress test their rigs can't be certain if their rigs are producing errors or not, and often they don't find out until their systems data is corrupted beyond recovery. This means that everything that is in main memory or is being written back to the hard drive could contain errors, and errors.

Prime95 should be ran more than once (I ran it twice} with a least one run of Short FFTs. Now its true that some computers that pass prime95 might crash performing a different set of tasks such as folding or playing a very graphically demanding game. When I was beta testing for FAH@home there were some teams complaining about not receiving points for completed Gromacs, only to be given the message that their Gromacs were corrupted. It is possible for a prime stable system to crash when playing a game because ones graphics card might fail through over heating, but Graphics card weren't designed to work with data.

 

Indeed, I'm pretty sure there were comments about folding at home causing errors when all was well with Orthos and OCCT. You just have to keep your eyes open when overclocking, in all areas. As for the CNPS9700, I'm not putting it down as a cooler, but even an Ultra-120 gets very hot when you run a 3.6Ghz quad with one, so a CNPS9700 will also get hot.