• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

i7 920, CPU vcore and dram raised, what next to make 4GHz stable?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
I had problems with my UD3R. I could not it get it stable @ 3.2Ghz much less 4.0Ghz. Didn't matter what bios, voltages, timings, cooling, etc.

I was forced to run stock speed until I swapped back to my Rampage III Gene.

No sooner I swapped I ramped straight up to 4.0Ghz right off the starting gate and I was back in business.

Don't get me wrong, the UD3R is a fine board, but not for me.

Good luck getting yours to 4.0Ghz.

I swapped to the UD3R because the PCIE slots are father apart on it than they are on my Gene which is a mATX. But it didn't work out for me so I went back to the Gene.
 
Hmm ok raised my IOH core and ICH core to 1.2

Ran OCCT for a bit, BSoD.

Here's the dump if it matters

It seems raising anything other than vcore causes BSoD's, since before this it wouldn't bsod, just fail a test.
 

Attachments

  • 012512-29796-01.zip
    19.9 KB · Views: 26
Last edited:
If changing voltage settings results in a BSOD. Either the board is board OR...

Auto setting is actually using a much higher voltage than what it is listing, so what you think you are increasing is actually decreasing.

Sounds to me like you need a better board though.

And yes IOH core 1.1 is Northbridge. IOH Core 1.5 is PCIE which is helpful too.
 
try
vcore: 1.3v
vtt 1.225v or 1.2
blck 180
multi 20

if it works, move up the blck and watch the vcore
i would reset the bios to all default settings before thou

and what neuro said
 
When overclocking I found that boosting PCIE voltage helped stability and actually improves scores in synthetic benchmarks. (IE 3dMark style benches) I would typically run 1.6-1.625 when benching.

thats what IOH 1.5 (or IOH PCIE) does.

Other boards call it different things as well.

As far as VTT (QPI voltage) you should have no problems running up to 1.35v for testing stuff out.

EDIT: I just realized you are running 20x multi. That can be troublesome. if you cant set 21.. leave turbo on

EDIT EDIT: I also just noticed that your board lists IOH 1.5 as "PCIE" that is what I was talking about earlier. Not important now, but after you get teh CPU stable...
 
Last edited:
Try odd numbers for your oc. I run 21*191. For some reason Nehalem like odd numbers.. not saying you can't get stable on even. But can't hurt to try!
 
i run a 930 at 21 x 191 on revision 1 of that X58A-UD3R. could have tried raising the uncore to x18 but someone might have mentioned that. Mine was really tough to stabilize at anything but x18 uncore
 
Okay.. reading up more on your board.

Performance enhance : disabled/standard
XMP: disabled

If you can not set 21x with perofrmance enhance off then leave it on, but do not allow the board to automatically set your memory speed and timings.


You can use the XMP profile to get the tmiings for your board at rated speed (in this case 1600) write them down.

then manually enter them, but drop mem multiplier to x6 so that we the most head room.
Drop uncore to x12 and QPI to 4.8GTs

Vcore - 1.35
QPI/VTT 1.3
Dram 1.65
PCIE 1.55
ICH 1.15
ioh 1.25
QPI PLL 1.15 raise this as Bclks increase to improve stability I needed 1.

(ICH IO you can leave at 1.5)


QPI PLL helps stabilize higher bus clocks. When really pushing it I keep it even with VTT.

190 was my limit on the bloodrage with out upping QPI PLL (called ioh VCCA 1.1 on that board )

If you still have issues try setting 101 on the PCIE. For 200+ you might need a little more, for 220+ you may need a lot more :) (someguys run 110-115 MHz for benching)

Once you have reached stability at 4GHz we can then focus on dropping some voltage or push a little further :)
 
Okay.. reading up more on your board.

Performance enhance : disabled/standard
XMP: disabled

If you can not set 21x with perofrmance enhance off then leave it on, but do not allow the board to automatically set your memory speed and timings.


You can use the XMP profile to get the tmiings for your board at rated speed (in this case 1600) write them down.

then manually enter them, but drop mem multiplier to x6 so that we the most head room.
Drop uncore to x12 and QPI to 4.8GTs

Vcore - 1.35
QPI/VTT 1.3
Dram 1.65
PCIE 1.55
ICH 1.15
ioh 1.25
QPI PLL 1.15 raise this as Bclks increase to improve stability I needed 1.

(ICH IO you can leave at 1.5)


QPI PLL helps stabilize higher bus clocks. When really pushing it I keep it even with VTT.

190 was my limit on the bloodrage with out upping QPI PLL (called ioh VCCA 1.1 on that board )

If you still have issues try setting 101 on the PCIE. For 200+ you might need a little more, for 220+ you may need a lot more :) (someguys run 110-115 MHz for benching)

Once you have reached stability at 4GHz we can then focus on dropping some voltage or push a little further :)

OK will try, but I know I can't get 21x without turbo on. QPI can't go lower than 36 for me except slow mode, so 4.8 is not an option.
 
Ok at these voltages I got a BSoD in Prime95:

vkG6S.jpg.png

I'm still running at 20x multi wth 200 bclk

The error code was 0x1e which people say is to raise vcore, should I do it? Doesn't seem like a D0 would need more than 1.35.
 
A D0 shouldn't need more than 1.3v for 4ghz to be honest.. Mine runs 1.25v. But I've got a pretty good chip.. I can run stock settings at like 1.1-1.125v.

Try 1.3v.

Also try 21x191.. Is your RAM rated for 1.5 or 1.65v? Leave it at its rated.. Though 1.65v shouldn't hurt it.. I see no reason to bump it up from 1.5v if that's what it's rated for.
 
A D0 shouldn't need more than 1.3v for 4ghz to be honest.. Mine runs 1.25v. But I've got a pretty good chip.. I can run stock settings at like 1.1-1.125v.

Try 1.3v.

Also try 21x191.. Is your RAM rated for 1.5 or 1.65v? Leave it at its rated.. Though 1.65v shouldn't hurt it.. I see no reason to bump it up from 1.5v if that's what it's rated for.

ok I guess I will trt 21x191, but won't that revert it to 20x during stress tests?
 
Back