K8N Neo2 suffers some SERIOUS VDIMM issues?

I was just tweaking my system up and I tried desperately achieve higher clock speed to my CPU, tried kinda everything out I could, more Vcore, more Vagp/Vchip, more vdimm, looser memory timings, dividers, all this without result.

After a while, I got my CPU running rock stable at 2520MHz with 1,55V (3500+ NewCastle, 252x10), Memory voltage(VDIMM) 2,65V and I was using 5:4 divider so my ram was running with 3-3-3-5 timings at 210MHz(Max Memclock: 166 used), VAGP/Vchip 1,50V. This far, all prime stable, superpi stable, 3d stable, etc, no problems at all.

So, I decided to go 2530MHz and higher. I fired my HTT (FSB) to 253MHz and it booted up. Went fine, and after Windows started I first ran SuperPi8M, 3DMark05 and PCMark04, it all went thru fine. Then Prime95, gave me error after 16mins. I just thought that maybe it didn't get enough juice, so I gave some more. I gave like 1,6V to cpu and I thought it should be stable now, but just a 8 minutes after I started Prime, I faced this same error again. Back to BIOS and I lowered my HT Frequency to 3x and I tried again, and now it took 11 minutes and Prime failed. Only problem I could imagine now was my memory, so I used loose 3-4-4-8 timings and guess what, Prime gave me error after 2minutes. I was totally pissed off and I thought CPU can't go any further.

This was rock stable:
CPU: 2520MHz (252x10 / 1,55V)
RAM: 210MHz (3-3-3-5 / 2,65V)
VAGP: 1,5V

And this was not stable:
CPU: 2530MHz (253x10 / 1,6V)
RAM: 210,83 (3-3-3-5 / 2,65V)
VAGP: 1,5V

And now comes the funny part. After like 20th visit to BIOS I increased my VDIMM to 2,7V and this wasn't stable neither, it was like 10 minutes and Prime failed. Okey, so AGAIN, back to BIOS and I ACCIDENTALY decreased my VDIMM to 2,6V and... it was stable, I ran prime like 6 hours and 59 mins and it was all ok. After that I was like "wtf I did?" I just decreased my VDIMM and now I'm able to do higher overclock. My system is now running 2550MHz (255x10) with 1,6V

So my stable system now looks like this:
CPU: 2550MHz (255x10 / 1,6V)
RAM: 212,5MHz (3-3-3-5 / 2,6V)
VAGP: 1,5V

This is more like a notice but I still got couple questions:
Like everyone knows, giving more overvoltage USUALLY gives you higher overclock, but what is happening here? I just don't get it.
And which causes this? A64's integrated memory controller, mobo, ram? I just don't know what to think.

Over and out

I found the same thing! my current rock stable settings:
FSB: 253
mult 10.5x
HTT 4x
mem 200Mhz 2.5-3-3-7
Vcore 1.525V
Vdimm 2.75V
PRIME95 48hours
it'll keep going unless I stop it...

Weird ehh it mystified me too....

Mine has done this as well. For instance when I set the vdimm to auto I can get through memtest at higher 1:1 setting than I can with any other voltage setting. Increasing voltage on my memory just seems to create errors in memtest and windows. This has not been the behavior of these sticks in the past.

Hmm, I don't know what the f--- is wrong with this. Also when you select Cell Menu -> DRAM Configuration -> RAS to CAS (Trcd) and change the value to "2" it seems like it forces voltage to 2,7V, no matter what you select from Cell Menu.

I wonder if MSI is aware of these kind of problems?
I'll wrote something to them so we can see can they fix this with new BIOS or something

Which BIOS version are you running anyway?

1.4 Official

me = 1.51b

Interesting.

You guys are using OCZ mem, I'm using Patriot. Now according to Patriot spec sheet Higher Vdimm is necessary. But now I'm curious. I got a rock stable 282 out of my memory with 2.5-4-4-10 so maybe.....

Yeh Neo2 does seem to have some voltage issues all around at least when it comes to stability testing.

Interesting....gonna have to take a look at this

sonicDivx said:

Interesting.

You guys are using OCZ mem, I'm using Patriot. Now according to Patriot spec sheet Higher Vdimm is necessary. But now I'm curious. I got a rock stable 282 out of my memory with 2.5-4-4-10 so maybe.....

Yeh Neo2 does seem to have some voltage issues all around at least when it comes to stability testing.

Click to expand...

I'm not using OCZ memory. I got my Infineon PC2700 stick from old Fujitsu-Siemens PC system and with my old P4 mobo I was able to push this as far as 260MHz (DDR520 / 3-4-4-8) with low 2,5V voltage, so I know personally my ram is not holding me on my overclock. I also tried to give more juice to it, but I did not gain any extra MHz so it was not worth of it.


Now we just need to find someone with OCZ DDR Booster to test out how this affects when VDIMM is raised thru Booster. Any volunteers?

No, I don't have a OCZ RAM Booster, sorry. I can tell you that with my RAM if I don't give it 2.85v I don't get out of the driveway. My system is at 240X9 right now and if I lower the vdimm to even 2.80v I won't even POST!

Let me update this; I did same thread to MSI forums and welcomes was not warm at all.

Link to MSI Forum & Thread Here

Not exactly the most usefull replies ever to be sure. You got an overclocking 101 lesson and the usual "that sucks I get better than that" stuff.

I think you're on to something this board definately seems to behave strangely to vdimm increases. I cannot find any settings that will make my OCZ PC4000EL stable at 1:1 in dual channel over 250Mhz. Kind of strange I cannot get even 1 Mhz beyond spec with OCZ memory regardless of voltage or timings. I've been fiddling with the 3.3v rail on my sparkle/fortron PSU but that didn't seem to do much so I added an additional AT PSU to run the vid card secondary power, HDs, & fans. That does seem to have improved my max stable OC somewhat though I still cannot get any more out of this memory.

At any rate it's a fun board to tinker with even if it is a bit quirky.

Zebbo,

Sorry to hear about your experience so far over there. Yeah this board really reacts to cpu and memory. I'm on another thread and this other guy gets 289Mhz 1:1 with the same memory I have, ack. So yeah there def. seems to be some factors at play.

Look forward to Nforce4 boards. Supposedly more stable with A64 chips.

Bro ive tried to recreate your problem as best I could but have been unable to do so. My guess is that only a few have that issue. Maybe its specific to the 54G?

sonicDivx said:

Zebbo,

Sorry to hear about your experience so far over there. Yeah this board really reacts to cpu and memory. I'm on another thread and this other guy gets 289Mhz 1:1 with the same memory I have, ack. So yeah there def. seems to be some factors at play.

Look forward to Nforce4 boards. Supposedly more stable with A64 chips.

Click to expand...

Thank you

But overally I'm damn glad about this board, this VDIMM problem just drives me crazy cause now I can't be sure what my CPU can and what it can't handle. I just hope MSI works this issue as fast as they can.

But this far 340x7,5 = 2550MHz / 1,55V Prime Stable so should I be something else then glad?

I haven't actually even started my overclocking yet cause I'm running stock cooling still (except I have Zalman NB47-J on SouthBridge). Let's see where I end after I get TwinMOS SP and XP-120.

Sentential said:

Bro ive tried to recreate your problem as best I could but have been unable to do so. My guess is that only a few have that issue. Maybe its specific to the 54G?

Click to expand...

Just a couple posts upward please. I am not the only one with this problem so I don't point this only at me. This far we don't still know what is causing this, and that's what atleast I'd like to know.

hmmm..

I'm having some memory issues myself with a NEO2. It just keeps getting more and more strange
Check the thread here:
http://www.ocforums.com/showthread.php?t=356367

If anyone has any suggestions I'd appreciate it....

Please, test this out and give report about your experience.

im about to order some ocz or ppxbl....i work at a computer store so i can swap the board if it blows ...i was goin to get the ocz membooster also....im assuming that the ocz willl run better with the booster...but i have some patriot bh5 now at it loves the voltages....so idunno...what do you guys think....