All I know is that I went from an instant BSOD, to a flash of the windows loading screen and then a BSOD after pumping up the FSB Termination voltage...so yeah it seems to do a tad bit of something
__________________
Asus Maximus II Formula
Intel Core2 Duo E8400 E0 Q828A Malay @ 4.50GHz (8.5 X 530mhz) WC'ed
2 x 2GB Mushkin XP2-8500 @ 1060MHz (5-5-5-15)
Asus 4870X2 @ 850/950MHz WC'ed
150gb Western Digital Velociraptor | 2 x 320gb Western Digital Caviar SE16 sataII
SoundBlaster Audigy4 | Logitech Z-5500 | LG W2600HP 26" TFT
MountainMods UFO | Zalman ZM-850HP psu
Watercooling setup:
1st loop -> Radiator: ThermoChill PA 120.3 | Pump: Laing DDC Ultra with Alphacool top | CPU: Swiftech Apogee GTX CU |Chipset: Swiftech MCW-NBMAX | Tubing: PrimoFlex 1/2" Pro LRT
2nd loop -> Radiator: Swiftech MCR-220 | Pump: Laing DDC Pro with Alphacool top |GPU: EK FC-4870X2| Tubing: PrimoFlex 1/2" Pro LRT
source -> line -> component -> short line -> termination
the "lines" in this case are signal traces. works a lot like a SCSI bus. high speed switching operations can cause reflections on the line (in essence, this is resonant noise caused from high-speed switching on the lines as the memory controller gates on and off to place data on the bus, nothing more than rising and falling voltages). clearly enough, the faster the signal switches (higher frequency memory) the more noise.
termination resistors, used in passive termination environments, are exactly matched to the target frequency for operation based largely on trace length, ESR (equivalent series resistance) and the over capactive/inductive nature of the circuit, as well as some other things well out of the scope of this post. problem is that this is often matched for a specific frequency...which as we know is not so cool for overclockers. enter the active termination system. this actually uses a voltage bias to set final line resistance value so that the user can tune for better operation when overclocking.
this method of tuning is largely trial and error as you have no way of know the specification of the circuit or have any idea of the calculations need to find the "perfect" value. so anway, the answer is, strangely enough, whatever works best.
Im sorry to bump up this old thread, but is there a limit whichi should stay within, or can i just rape my Wolf with 1.5-1.6V FSB termination voltate?
I believe for the 45nm parts Intel has a safe range of 1.045V < 1.10V <1.155V (higher for 65nm parts) but there are those that are running as high as 1.30V--I am running @ 1.16V for 400FSB. If you go to their sight you should be able to find the data sheet pertaining to your chip that gives the safe limits for both vtt and PLL.
45nm = 1.15v max, 1.45v absolute max. Going above the max may immediately cause irreparable damage, and the functionality nor reliability of the CPU can no longer be guaranteed. I prefer to stay under the stated 'max' of 1.15v.
45nm = 1.15v max, 1.45v absolute max. Going above the max may immediately cause irreparable damage, and the functionality nor reliability of the CPU can no longer be guaranteed. I prefer to stay under the stated 'max' of 1.15v.
As far as I'm concerned, the default FSB Termination Voltage by Intel is:
1.1V for the 45nm parts
1.2V for the 65nm parts
As for what's safe and what's not... the maximum voltage of 1.45V doesn't seem to be the real maximum anyways ( at least 3 of my Core 2 Duo's 45nm disagree ) [ and I will be putting even more CPUs to the test, and even higher voltages ]
__________________
Many thanks to: Sue Wu, Yiwen Lin, Steven Kuo, Vivian Lien, Joe Chan, Sascha Krohn, Joe James, Amy Deng, Jack Peterson, Hank Peng, Mafalda Cogliani, Marta Piccoli, Oliver Baltuch, Steffen Eisentein
should 1.2vtt be considered high then? what about a p45 NB whats considered high over 1.3v ??
I am getting 518x8.5 stable on a e8600 and it takes 1.2vtt and 1.26 on the NB.
I really don't see it being necessary to exceed 1.3V on the NB for the p45 chipset. I hit 450FSB using 4x1GB of RAM >1100MHz on a quadcore w/ only 1.24V, so unless you are just seeing how high you can go for benching >1.30V shouldn't be needed. The same also applies to vtt, I don't really see why 1.2V would be necessary for a 24 hour clock IMHO.
Just remember there is also such a thing as TOO high. Setting the FSB VTT to 1.4 assuming that'll cover you for any kind of overclock is wrong.
There is too low and too high. And I don't think anyone needs 1.4 as I can do 4.25ghz on 1.2V so .. unless you're on LN2, start with 1.2 and work slightly up/down from there
45nm = 1.15v max, 1.45v absolute max. Going above the max may immediately cause irreparable damage, and the functionality nor reliability of the CPU can no longer be guaranteed. I prefer to stay under the stated 'max' of 1.15v.
thats the reference VTT not the max, the max is usually near the max rating for the core
but intel no longer lists it on their spec sheets
__________________
warning i use sarcasm, alot
coming soon duel dual rad p180 the reckoning, 720BE 0912epmw, gd70 and a new liquid revamp
give a man some juice and he has juice for a day, give him a juicer and he will have juice for life if he has something to juice
Linear Mode
