DFI NF4 Thread

[Blue078]

Originally Posted by Bampson

I posted a while ago about getting stable at stock settings. Long story short, I have achieved that now, but oddly enough I had to adjust the vdimm down to 2.7v from what I was using previously.

Initially I was using 2.8v or even 2.9v as suggested by many others for the TCCD sticks in dual channel to "get up and running". But after fiddling with many settings, I determined that it was this setting alone that was causing instability.

So you may be thinking just keep the setting down and be happy, but the problem is that these sticks should handle 2.9v no sweat, and I would like to have the headroom if and when I overclock.

The other odd thing is that memtest86 will pass 20 some runs at the higher voltage, but windows craps out before hitting the desktop frequently until I back it off to 2.7v again.

OCZ has been kind enough to offer an RMA if needed, but now I'm wondering could this be the memory controller on the CPU or even the DFI board itself?

Which BIOS are you useing ?
You know alot of the early TCCD chips didnt like higher voltage than 2.7.

[xgman]

Originally Posted by Blue078

I can get 312 out of my Plat rev2's with the 0209 BIOS 2-4-3-7-1t.

so it's not the board then. doesn't 312 2-4-3-7-1T give you higher bandwidth than 2-2-2- at 250htt? (assuming no divider)

[xgman]

Originally Posted by Bampson

I posted a while ago about getting stable at stock settings. Long story short, I have achieved that now, but oddly enough I had to adjust the vdimm down to 2.7v from what I was using previously.

Initially I was using 2.8v or even 2.9v as suggested by many others for the TCCD sticks in dual channel to "get up and running". But after fiddling with many settings, I determined that it was this setting alone that was causing instability.

So you may be thinking just keep the setting down and be happy, but the problem is that these sticks should handle 2.9v no sweat, and I would like to have the headroom if and when I overclock.

The other odd thing is that memtest86 will pass 20 some runs at the higher voltage, but windows craps out before hitting the desktop frequently until I back it off to 2.7v again.

OCZ has been kind enough to offer an RMA if needed, but now I'm wondering could this be the memory controller on the CPU or even the DFI board itself?

Take the RMA. Newer tccd sticks handle up to 3.1v. Mine run nicely at 300htt set at 3.01v.

[Blue078]

Originally Posted by xgman

so it's not the board then. doesn't 312 2-4-3-7-1T give you higher bandwidth than 2-2-2- at 250htt? (assuming no divider)

Yes Sir it show doos
But my next CPU will run higher multi's so I would like to get some ram that can do all the 2's at 250 to 260 and still be able to max my chip out.

I mainly bought the Mushkin Blue because its $240 for 2 gigs, Now I can pull my BH-5 out of my server and use it for play .

[Bampson]

Originally Posted by Blue078

Which BIOS are you useing ?
You know alot of the early TCCD chips didnt like higher voltage than 2.7.

Using the newest 3/10 Bios. These may indeed be older chips even though I just ordered from Newegg last month. Could be they've been sitting in inventory for some time.

In any case I think I'll RMA and try out some hopefully newer yields. Thanks for the tip.

[Blue078]

Originally Posted by xgman

so it's not the board then. doesn't 312 2-4-3-7-1T give you higher bandwidth than 2-2-2- at 250htt? (assuming no divider)

Well actuly I dont count 318 because HL2 will crash sometime within 30mins. But I have ran a bench or 2 at 318 on 2.9vdimm.

You remember Garry got 320 easy as 1-2-3 with the PQI TCCD

[Eldonko]

My TCCD shows no difference in performance from 2.8-3.1v and is useless at 2.6 or 2.7. Guess it is newer stuff.

[Sierra_abx]

If the Samsung chips are marked "0440" or later, then they probably can take higher voltage. I have a pair of G.Skill LA sticks (0440) that can even run at 3.4v. Will probably sell them since I have no real need for them.

[eva2000]

Originally Posted by Blue078

Heres the deal, I have 3 gigs of Twinmos BH-5, I have tried 3gigs of VX 3200, and now 1 gig of Mushkin blue.

I could get all of them to run 240, it over 240 that is makeing me "CRAZY " That is just to many sticks to be limited to 240 or under. Logic says Blue078 is doing something wrong..... Fate says he my die before he figures it out

woah alot of testing but at what volts ?

[Blue078]

Oh! I see why Seirra likes the 217 BIOS I still get errs in mem test8 at 250 with the MushBlue,and BH-5, but not as many {5 ever pass} at 3.4vdimm.

It has got to be some timing setting I'm over looking

[Sierra_abx]

Your Mushkins may just need time to break in. 250 is better than what many others are getting. My Level 2's get Memtest errors at 260-270 with 3.5-3.6v. But I haven't used the sticks that much.

[Blue078]

This 217 BIOS is kick *****

Drivers set to high quality no tweeks, drive image/windowblinds/all services runing in the background.

[Sierra_abx]

Originally Posted by Blue078

This 217 BIOS is kick *****

See, you should listen to me sometimes.

[Blue078]

Originally Posted by Sierra

See, you should listen to me sometimes.

But but but It didnt work at all for TCCD, for me anyway. But it shows doos wounders for the UTT, and "REAL" BH-5.

Now for the most important test, how long can HL2 last with out issue ???

[xgman]

Originally Posted by Blue078

This 217 BIOS is kick *****

Drivers set to high quality no tweeks, drive image/windowblinds/all services runing in the background.

Ther you go.