E8400 Overclocking On DFI Blood Iron

[mattwalter85]

I have to run mine at 1.55V to run 4.3Ghz stable which may be a little higher then u'd like to take it.....

[jammin]

Yeah, that's about .2v higher than I like to take it.

 

I might end up going to 1.4v, but I'm going to try and exhuast tweaking other settings before going there.

[mattwalter85]

I took mine to 1.6 one time but the performance diff. was minimal so i went back down to 4.3 from 4.7Ghz

[capthowdy575]

I took mine to 1.6 one time but the performance diff. was minimal so i went back down to 4.3 from 4.7Ghz

man i would hate to be your cpu. but nice oc on the 8400 was thinking about picking one up.

[jammin]

Right, I have been doing some further tinkering this evening, mainly to test out some of the other voltage options to see how they affect stability.

The answer so far is not very much.

I've played a little with the NB, VTT, GTL and Clockgen Voltages, but they don't seem to have increased stability at already unstable speeds.

Haven't touched the SB Core/CPU PLL Voltage yet.

 

Anyway, I'm still at 3.6GHz for when I'm not testing at the moment, though with some slight improvement in RAM settings.

 

Settings for 3.6GHz are currently:

 

450MHz x 8 multi

 

RAM Timings: 4-4-4-12-2T @ 1.95v

 

Voltages in BIOS:

 

CPU Add : +100.0Mv (so CPU= 1.3v)

SB 1.05: 1.070v

SB Core/CPU PLL: 1.55v

NB Core: 1.33v

CPU VTT: 1.20v

CPU Clockgen: 3.60v

 

All GTLs @ 0.67

 

 

The highest I have managed to get reasonably stable (1hr+ Of Orthos before stopping without a crash) is 3720MHz @ 1.34v

 

Will keep plodding on, but it is looking like the CPU will need more voltage to go further. 4GHz+ isn't looking too likely for a 24/7 clock at the moment.

Edited February 25, 2008 by jammin

[Shadowfactor]

your box is different then mine. Like the label looks a lot different and it doesn't even have a batch code.

some of the E8400 have actually turned out to be OC dud's. Before I exchanged my first chip, I couldn't get over 3.7 ghz truly stable even with 1.55V

Which had a production date one day later than yours. ( 1/10/08 )

 

The new chip I have so far really likes anything higher than 3.8 ghz. I can't even get a stable oc at 3.2 ghz. Stable OC's don't happen to 3.8 ghz.

I haven't really tried any farther than 3.9 ghz currently as I just haven't had the time.

 

When I was reading about OC'ing and effect of NB and SB voltages. It seems to be that a good amount more of stability can be achieved from cooler NB.

People were reporting that if they cooled there NB from 50ish C to 40ish C that they could achieve a good extra 200-300 mhz.

SB voltages are also important for stability, so try maybe bumping it up a little and see what that gives you.

 

Also your NB voltage seems really low. If I can remember correctly, most users running Oc'ed P35 boards were giving the core around 1.45 V.

[jammin]

Batch # is Q744A761 (it is on the box ).

 

I haven't pushed anything except the CPU voltages higher than a couple of extra notches from default so far, and have been playing with them individually rather than at the same time.

I'm trying to be methodical with testing, which while the best approach in my opinion does get tedious rather quickly.

With all that rebooting, changing one setting, running Orthos.. crashing, rebooting, bumping it up another notch, running Orthos.. crashing, rebooting, changing a different setting... you see where this is going.

 

Basically it takes a long time to test any of the options to see how much affect (if any) they are having.

Especially if you factor in multiple Orthos runs each time (it doesn't always crash at a consistent point).

 

I'm confident that I'll eventually get a bit further without pushing too much voltage through the CPU, but it will be slow and steady progress.

I haven't even tested for bench stable settings yet, but they are of course likely to be higher.

[razor]

jammin,

 

I'm reading up on methods other than brute force vCore/VTT settings, and many of the more experienced guys on the DFI boards have alluded to the necessity of learning the GTL settings in relation to VTT, vNB, etc. allegedly, tuning w/GTL settings can help acheive/stabilize an OC w/out dumping tons of voltage into the chip. at least, that's how I'm understanding it. I'm sure you've already seen it, but just in case - here's a link:

 

- Blood Iron sticky @ DFI Club

 

the GTL thing seems to be very important to OC-ing C2Q's, and I noticed that the Blood Iron BIOS looked quite similar to the LT-P35-T2R's. here's a link to an Anandtech article pertaining to OC-ing the LT-P35-T2R, giving parameters for GTL tuning, VTT for FSB ranges, etc. hopefully it'll help you out

 

EDIT - I know you're not using a C2Q, but I cant imagine that fine-tuning these settings would have no positive effect on your setup

[jammin]

That's interesting, because the LT-P35-T2R allows independent adjustment of GTLs while the Blood Iron gives you values linked to the CPU VTT like this:

 

x CPU GTL0/2 REF Volt.......... - 0.67X

x CPU GTL 1/3 REF Volt......... - 0.67X

x North Bridge GTL REF Volt.... - 0.67X

 

Which would give you 67% of the CPU VTT as I understand it.

Other values available are something like 0.63, 0.61 and further down, but not above 0.67.

 

I'll play with them a little more, but testing adjusting them individually (Orthos runs) didn't show much of a difference to stability.

I'm starting to get a hold on the theory though, which is handy (I don't like adjusting things I don't understand fully).

Edited February 27, 2008 by jammin

[razor]

I'm starting to get a hold on the theory though, which is handy (I don't like adjusting things I don't understand fully).

 

 

I'm the same way...that Anandtech article has helped a ton, as has DFI Club. another new-ish site is Edge of Stability, which goes pretty deep (for me) into the mathematical side of GTL tuning. might be worth a look for ya.

[Bleeble]

Am I the only one that thinks DFI goes out of their way to make things needlessly difficult? Anyone else notice how many DFI boards are getting world record overclocks?

[razor]

Am I the only one that thinks DFI goes out of their way to make things needlessly difficult? Anyone else notice how many DFI boards are getting world record overclocks?

 

I'll certainly admit that this is easily the most complex board I've ever spent time with, but I guess if your company wants to be known as the choice of the hardcore enthusiast...and honestly, I like learning about the intricacies.

 

something I've wondered while I'm fiddling w/this board is whether DFI makes the BIOS so detailed because they feel that

 

a - they feel that the "simplified" settings of other BIOS adjust values too coarsely

or

b - they feel that bludgeoning your CPU w/voltage is boorish

 

there's also a strong possibility I'm just talking out of my .