Oops! P4 2.6 SL6WS won't overclock?

[noskill]

Guys,

I have a problem here.

My rig:

P4 2.6 (SL6WS) @ 3.145 (242 x 13), Vcore = 1.7V, D1 stepping, Malay
ASUS P4C800 Gold
Kingston ValueRAM DDR400 3-3-3 (works at 2-3-3, he-he) 512 MB x 2

I'm using Arctic Alumina and Thermaltake Spark7+ Cu on CPU. Temps are somewhere near 43-45 in load.

The problem is that CPU won't overclock any more! I thought that memory (c'mon guys, that's usual ValueRAM!), but then set divider to 1.6 (i.e. 1.6 x 242 = 387 MHz), and timings to 3-3-3-8, but that doesn't change anything.

When I set CPU somewhere near 3.3 or even 3.4, it simply either doesn't load Windows or just hangs in Sandra. It doesn't hot too much - 43-45 in full load.

Flashed BIOS from 1.011 to 1.014 and 1.010 -- that changed absolutely nothing.

I've read various forums, where exactly this CPU was overclocked stable at 3.4-3.5. Some ub3r_hax0rz pulled it to 3.7! (with waterchiller, of course).

So -- can you reccomend anything concerning this problem?

[macci]

've read various forums, where exactly this CPU was overclocked stable at 3.4-3.5.
That doesn't mean that yours will go that high. There are good chips and bad chips. Some clock to 3.4 some dont. Simple as that.

[macci]

It is possible that the RAM has problems doing 5:4 mode too thou. See if it has 'CH-5' chips on it. That might be your problem too.

Does increasing Vcore make any difference? Does it clock to the same speed at Default Vcore and at 1.7V?

[noskill]

Originally posted by macci
That doesn't mean that yours will go that high. There are good chips and bad chips. Some clock to 3.4 some dont. Simple as that.

Of course -- but they do have exactly the same chip revision!


It is possible that the RAM has problems doing 5:4 mode too thou. See if it has 'CH-5' chips on it. That might be your problem too.

Does increasing Vcore make any difference? Does it clock to the same speed at Default Vcore and at 1.7V?

Thanks, I look at it when get back home.

But still can't understand why it might have any problems. If we use 1:2 mode and set 200 FSB, memory work at 400 MHz. If we use, say, 1:1.6 mode, then at 250 FSB memory should work @ 400 MHz. So why it might haveany problems with 5/4 or any other dividers?

BTW. Take a look at this page, it describes my latest achivements in overclocking this rig: http://www.dubina.ru/default.asp?id=tx-0

You see, I was forced to lower FSB, because it freezed somethimes when Folding for about 36+ hours. But Sandra and 3DMark passed successfully in a 10 loops each!

Now to Vcore. Yes, it does matter. For example, if I set Vcore at 1.65 with 239 FSB, it fails. However, it works strange. I thought raising Vcore will result in temperature increase. So I'm able to set 1.95 with my mobo. So I've done at set 248 FSB. It doesn't work at all! Then, I change Vcore with -0.025 each time and it doesn't even show POST before I reach somewhere at 1.75. Neither Vcore doesn't work. Higher Vcore results in not passing POST, and lower -- it fails tests.

I thought that's bad cooling, but temperatures are somewhere at 53-56 in full load.

[macci]

Of course -- but they do have exactly the same chip revision!
Same rev.?
Like same CPU prod. code? or same FPO/Batch?
And even if those details are the same it doesn't mean yours will clock the same! There can be a difference of 200-300MHz easily.
Since it runs better with higher Vcore I'm quite sure the chip is just a poor clocker.

I was thinking about a mem problem since there are CH5 based modules which will not go above 250-260FSB no matter what RAM divider or settings are used. Its a problem with those chips and i865/875 platform. But I guess this is not the problem here.

[Bravo]

I have the same stepping, and speed processor. Its not good for much over 3.25ghz @ 1.6v :(

[noskill]

Originally posted by macci
Same rev.?
Like same CPU prod. code? or same FPO/Batch?

I mean that guys also have SL6WS processor...



And even if those details are the same it doesn't mean yours will clock the same! There can be a difference of 200-300MHz easily.
Since it runs better with higher Vcore I'm quite sure the chip is just a poor clocker.

Could be this because of bad cooling? You see, when I left my home I set RPM on my TT Spark to maximum. When I came home in a few hours, it never freezes. Then I lower RPM, because I can't manage that noise! Then it might freeze in a hour or so, but temps are somewhere near 53-56.

I'm going to build Swiftech waterchill system. How do you think -- will it let me push CPU a bit more?

[bldegle2]

pure and simple.

if you go the thermaltake 947 route w/ a 92MM fan, you will see an improvement is both FSB and total MHZ as the temps drop. you will also reap the benefits of the higher voltages once you get the temps down, the cpu is just not liking the extra heat when 'juiced', and the massive 947 will draw tons more heat out faster.

i don't know how much more you will get, each cpu is different, but if you really want to max on air only, you got to upgrade the cooler.

good luck.

baldy

[Tedinde]

I got 3 of the same as yours, one wouldnt go over 224fsb!!! even in my vapo. but one would do 290.

Steppings and ID mean nothing nor how new or old the chip is. Best C chip i ever had was my first 2.4 that's 7 months old now.

[noskill]

Originally posted by bldegle2
i don't know how much more you will get, each cpu is different, but if you really want to max on air only, you got to upgrade the cooler.

I'm going to build a Swiftech-based waterchiller system. Hoping that would help...