Anyone know why I can reach much higher Bclks using the Asrock OC Tuner than
via the BIOS? With all other settings the same (VTT 1.429, PLL 1.853, PCH 1.1), I
can reach 214 Bclk using the OC Tuner, but only up to 198 doing the same thing
via the BIOS. In each case, I'm testing via (for the moment) 10 minutes of Prime95.
I expect I could reach more than 214 via the OC Tuner if I increased VTT, but 214
is plenty. Just wierd that I can't even get a POST if I use the BIOS to set the same Bclk.
CPU multi is just 12 atm as I'm only at the stage of finding the max Bclk. Vcore is
less than 1V (Auto). The usual stuff disabled (Vdroop, SpeedStep, CStates, etc.)
System is an Asrock P55 Deluxe, i7 860, U120E Rev B, 4GB GSkill Ripjaw DDR3/2000,
WD VR 150GB 10K, two SFF12Fs for case cooling, Sycthe Kaze-Jyuni 1900rpm for
the U120E, two ASUS 8800GT SLI, Thermaltake Toughpower 750W, running XP Pro SP3
32bit with Autopatcher.
I'm not sure why that would be, but some boards are just picky. I would up your PCH voltage a tad bit.
Also, make sure to keep VTT at 1.219v or lower, as that is Intel's absolute MAX (not the recommended max, even). The uncore of Lynnfield and Clarksdale I guess isn't as robust as the one on Bloomfield.
For reference on your problem, myself and a friend were able to get an old Pentium 4 (Prescott) to 4.5GHz using dry ice, but couldn't boot above that setting. In Windows, though, we were able to hit 5GHz without many problems.
> I'm not sure why that would be, but some boards are just picky. ...
I wondered if it was something to do with the possibility that, with the system already booted, it's easier for the
OC Tuner to make changes 'live', as opposed to cold starting straight into some kind of high-spec setting. I'll
probably find out more later when I've sorted out the max RAM speed and then finally the main CPU oc.
> ... I would up your PCH voltage a tad bit.
I tried that, didn't seem to make much difference.
> Also, make sure to keep VTT at 1.219v or lower, as that is Intel's absolute MAX (not the recommended max, even).
> The uncore of Lynnfield and Clarksdale I guess isn't as robust as the one on Bloomfield.
Nah, it doesn't need to be that low. From all that I've read, the main two rules of thumb seem to be:
VTT < default + 0.3
Vdimm < VTT + 0.5
I've checked many posts from people with oc'd 860s to see what settings they used, plenty have had VTTs of
1.4 to 1.45 without any apparent problems.
> For reference on your problem, myself and a friend were able to get an old Pentium 4 (Prescott) to 4.5GHz using dry
> ice, but couldn't boot above that setting. In Windows, though, we were able to hit 5GHz without many problems.
Hmm, it would be nice to have a setup which booted straight from the desired settings, but if the OC Tuner is able to
produce an oc config that's definitely stable then I guess that's ok.
I suppose it doesn't really matter either way though because even 198 is more than enough to reach my 4GHz target,
ie. 21 x 198 is 4158. However, if a higher Bclk was possible then I could use a lower multiplier to reach the same
CPU clock and thus run the RAM at a faster speed (since it's DDR3/2000); I have a second 4GB DDR3/2K kit I've not
tried yet aswell (Kingston HyperX).
Note that I've been following this guide more than any other:
but I've read a lot of articles/reviews and forum threads to seek wider opinions. I'd originally begun with the proverbial
blind stab in the dark approach (I know, unwise), which gave a stable 3975MHz oc, but I decided to step back, start
from scratch, do it properly. From reviews I've read though, I was expecting the BIOS approach to give better results
than using the OC Tuner.