ClockGen discrepancy

[Liquid3D]

I'm trying a 3.0C I traded my 2.4C, a 1GB-kit of Corsair Twin-X4000 memory for and there's absolutely no getting above 3824MHz or 254FSb. Now I'm not upset because when we traded he said it did about 3.6GHz, and 3.5GHz stable. So I expected less. That's not the issue however.
Trying to reach the 4GHz mark, I pumped as much as 1.74Vcore, and even had Vcore up to 1.8V the BIOS. The temps of course were below 21C or I wouldn't have pushed it. It's so frustrating because it posts into 4GHz, but the Window's splash screen won't come up.

Anyway I thought I try using ClockGen, and even tried PLLSoftFSB, and this is what occurred. CPUID would read it as 3.8GHz, and CPUZ would read it as 3.9GHz. Sandra also seemed to confirm, that the FSB had remained at 254FSB instead of 260FSB. Can anyone help me out here?

[saaya]

i wouldnt trust sandra and wcpuid, i think the clockgen and cpu-z are right...

have you tried the realtime clockspeed measurement on wcpuid? what does it say? check sandras memory bench to see if its faster after you used clockgen!

[shortcircuit]

I get the same thing when changing the fsb on my dually tyan tiger mpx, but if I reboot it, wcpuid starts showing the new speed. I don't think the speed increase takes effect until after the reboot (benchmarks confirmed this).

[Soulburner]

If your board has any Turbo mode that could be what is doing it. Turbo modes on some boards overclock the FSB a few for extra performance.

Also get CPU-Z 1.20a.

[macci]

Its all because of HT and WInXP. If you had installed WinXP without HT this problem would not occur.
Also if you select the 'realtimeclockchecker' from WCPUID and from the properties select that you are using 'SetFSB with Win2k/XP' it will display correct speed too.
Clockgen works and the MHz will actually go up but WCPUID, sandra etc just can't 'see' it. has something to do with HT.

[Liquid3D]

Thank's fellows. I can understand how this happens, I can't remeber what exactly is going on, but I beleive your right re-booting would change it, problem is when you re-boot the BIOS values become the "real" values, taking precedence over ClockGen. I've tried the "RealTimeClock" in WCPUID, and it reads the first FSB/Processor speed. I've tried Refreshing both CPUID, and the desktop.

My primary concern is it's not the other way around. What I mean is I'm more concerned that the "true" frequency is the 3911MHz. I don't necessarily have to becnhmark from it, I'm simply trying to determine the Processor's limits. I do beleive changes are occurring because when I try to take ClockGen or SoftFSB further to 4GHz, the system locks, and/or crashes. As I said what's important to me is that I'm finding the Processor's true potential.


I did find it hard to beleive that ClockGen would simply "trick" monitoring software into believing a certain frequency had been acheived. Can the Clock be "tricked" into reading a certain value which the software woukld read? On that note, what program did Kenneth Chong use to test his frequency? Could he have been the victim of ClockGen making a "front side adjustment" to the clock, but not acutally changing the Processor frequency?

[Soulburner]

I don't think it has anything to do with HT, at least my in case. I always install XP with HT on and I never have any problems with that, CPU-Z always recognizes my speed correctly when I use ClockGen.

[shortcircuit]

Originally posted by Liquid3D
On that note, what program did Kenneth Chong use to test his frequency?

LOL, I think he used Photoshop.

[cpulloverclock]

That's possible and even sure

[kromosto]

only cpuz can see the changes from setfsb or clockgen in my system too others sisoft or cpuid cant see them thats normal

[Liquid3D]

Although I shiver and shake even to mention this. This chip was run in a Vapochill for quite a while using PCM+, which of course is not recommended for phase-change. So I wonder if it was damaged? I can't seem to finish 3DMark2001 loop, at 250FSB, which should be like butta.

The memory I'm using now will run stable at 284FSB forever, and the Asus 9950 I have is in good shape? I did OC it but from 400/850 I only raised it to 423/876 (443/881 crashed). I upped the VDDQ but this wasn't necessary anyway, because it's run higher at default.

What I'm saying is I beleive the 3D-graphics "bus" on the Processor, may be damaged? Is this possible? Anyone know

Honestly who wouldn't try to run their 3.0C at 4.0GHz? I may be asking alot, as it seems the Northwod-D's are running as expected, with a 400Mhz to 600MHz performance ceiling over the highest model's clock speed. Therefore I can't fault Intel, but this still has me wondering why so many 3.2C's will reach 4.0GHz, while so many 2.4C's and 3.0C's will not? After all Intel doesn't condone overclocking anyway, so why would they hand-pick silicon capable of 4.0GHz, for their 3.2C's over their 3.0C's? Obviously the multipleir makes a difference, but does it really? I've had better overclock's striking a balance between multipplioer and FSB. I guess it's as simple as their preserving the best silicon for the highest models numbers, and what I have here is one that fell backwards through the binning cracks.