DFI Admits 5Volt Vdimm Problem

Printable View

Show 100 post(s) from this thread on one page

05-27-2005, 01:41 PM
pershoot

Quote:

Originally Posted by drunkenmaster

pershoot, run 3.5/3.6v , its less hard on the board, and most utt does keep scaling up to those voltages so you should get a bit more speed anyway. its safer imho.

i dont feel safe running above 3.4 24/7.
05-27-2005, 01:43 PM
hovo73

Quote:

Originally Posted by bachus_anonym

Am I the only one to notice, that folks with "old school" BH-5 DO NOT have this problem ???

Yes you are. Can all the fanboys please just stop and wait to see how this thing plays out? In the other thread on this forum somebody posted that his old BH-5 was worknig fine for years on another board with some very high voltage, but died on DFI. And lets not forget that some people with TCCD also had this problem. And please stop blindly blaming the user.

And one more thing, DFI most certainly used the 4V DIMM option to market the board. Back in November/December when most of the other NF4 boards were out people were waiting for DFI to come out exactly for this reason. One more time, it is a feature of the board, and I expect it to work as intended.

Once again, lets wait to see what we hear from DFI. I have to say though, the fact that DFI-Street mods are quite about this, and the fact that they are closing the threads about this left and right, doesn't inspire to much confidence.
05-27-2005, 01:46 PM
[XC] 4X4N

Quote:

Originally Posted by bigtoe

I have been taking most of the EU RMA's for OCZ for a while now, I have a huge amount of dead ram here, i talk with most of the users to see what they found etc and i can say this does NOT start and finish with UTT One guys has corsair and kingston BH5....killed one module from each kit. He has had 4kits of VX and 2 kits of twinmos BH5...all have had one module killed. This is just one guy....i have story's from many more.

Now i know you all love this board...i do to, issue is i don't like it killing memory, if it hasn't killed yours and you are on the 5v line, you ain't used it long enough yet....thats how I feel about the situation at the mo and im working VERYhard to help solve this problem.

All those people who are saying it has to be user error, go read on multiple forums to see what people are having issues, see how they set their rigs up and you wuill see they are just as thorough as you are and as I am.

Im going to say this again, come off the 5v line, tweak your 3.3v rail and enjoy your memory and your motherboard.

Hipro seems to think his maximizer works well and we have found the booster does also...maybe this could be another answer.

Thank you for all of your work on this. I am glad that there is someone who acknowledges that there could be a problem with the board. I love the board too, but would like to see this issue resolved. I also agree that is seems that this happens to users who run their computer more. Mine runs 24/7. Hardy ever is turned off. I have it set up like a htpc. The wife uses it every day like a tivo. I would volunteer to help in the cold boot thread, but I rma'd my bad twinmoss ram and haven't received the new stuff yet.
05-27-2005, 03:36 PM
Ouchy

Quote:

I don't. Many of us, including me, lost RAM while using the 5v feed even though we were only using 3.2volts and have active cooling (I had 2x80mm fans running directly on them. This is not on the user end as much as it is a glitch with the 5v feed.

I don't think you specified in our previous discusssion, but what was your PSU before you got that Powerstream? Was it 20 pin? I only ask because you seemed surprised when you linked to a post that says that a 24 pin PSU is required for use on the DFI board.
05-27-2005, 03:47 PM
ben805

ok, I've got a question regarding using the OCZ booster as a workground and possibly better solution than 5v rail, isn't the booster came with 20pin connector to "tap" onto the PSU? so what happen to those who own 24pin like the PCP&C? does the kit came with 24pin too? I'd much rather spend the 35bucks than crack open my $240 PSU and void the warranty...

isn't this a 20pin?
http://images10.newegg.com/productim...998-301-03.JPG
05-27-2005, 03:58 PM
muzz

Quote:

Originally Posted by Ouchy

I don't think you specified in our previous discusssion, but what was your PSU before you got that Powerstream? Was it 20 pin? I only ask because you seemed surprised when you linked to a post that says that a 24 pin PSU is required for use on the DFI board.

Did they specify that in the beginning/manual?
I dont remember that.
05-27-2005, 04:15 PM
Jessfm

Quote:

Originally Posted by muzz

Did they specify that in the beginning/manual?
I dont remember that.

It was officaly released information, afterwards IIRC
05-27-2005, 04:22 PM
Tony

Quote:

Originally Posted by ben805

ok, I've got a question regarding using the OCZ booster as a workground and possibly better solution than 5v rail, isn't the booster came with 20pin connector to "tap" onto the PSU? so what happen to those who own 24pin like the PCP&C? does the kit came with 24pin too? I'd much rather spend the 35bucks than crack open my $240 PSU and void the warranty...

isn't this a 20pin?
http://images10.newegg.com/productim...998-301-03.JPG

No 24 pin version as yet, with a Powerstream you just unhook the extra 4 pins and plug them in...next to the booster.

Now i know this does not help you but Im sure we can look into a 24pin version soon.
05-27-2005, 04:29 PM
s7e9h3n

Quote:

Originally Posted by bigtoe

No 24 pin version as yet, with a Powerstream you just unhook the extra 4 pins and plug them in...next to the booster.

Now i know this does not help you but Im sure we can look into a 24pin version soon.

So you're saying it's possible to use the booster on a dfi nf4 board? What's the extra 4 pins which you talk about on the powerstream?
05-27-2005, 05:13 PM
odb

Quote:

Originally Posted by s7e9h3n

So you're saying it's possible to use the booster on a dfi nf4 board? What's the extra 4 pins which you talk about on the powerstream?

its another 4 pin connector that slides on to the 20 pin power connector to make it a 24pin, mine got it, i was thinking about selling my booster after just buying this ultra-d but i might us it instead of the 5v jumper now.

just one question about the booster on the dfi board, i got my vx in the 2 orange slots, what configuration should i use to use the booster? which slots for the ram, and which slot for the booster?
05-27-2005, 05:22 PM
topboy

In order to deal with OCZ support questions and for tracking purposes, could you direct your questions over at the official OCZ forums at

http://www.bleedinedge.com/

You will get all the technical questions you need there very quickly

A OCZ picture on the right of the page will take you to the support forums

Thanks

Topboy
05-27-2005, 06:01 PM
deception``

Quote:

Originally Posted by muzz

Did they specify that in the beginning/manual?
I dont remember that.

The 24-pin requirement was mentioned by DFI long before the release of the NF4 motherboards.

deception``
05-27-2005, 07:50 PM
wtz54321

I have two stupid question ....

1 . How OCZ can feature 3.5V +- 5% spec for the VX series of memory ? :confused: We all know that there is no such memory chip from any vendor to provide this kind of spec for use in 7/24 enviroment ... Even the old BH5 can only handle the 3.6V absolutely maximum rating voltage , but this is absolutely maximum rating , not 7/24 working voltage guranteed ... I really think that this is some kind of risking business IMHO :nono: ...

2 . How can we be sure that the dead DIMM is really killed by 3.4V or 3.5V setting in the bios as they claimed ? Who knows what's the voltage that ever supply to the DIMM in it's lifetime of operation ??? :stick:
05-27-2005, 07:55 PM
Snark42

Quote:

Originally Posted by drunkenmaster

again, from reading a few forums, yes some people have issues, i've seen one thread, and then the guy big toe mentioned both people with like 10 sticks of dead ram. both of these guys have a set of mem die and keep throwing more sticks in not thinking maybe its the board. some boards of any brand will have a faulty part and can cause issues. no offence but if i put in 2 sticks of mem and they died, i probo would try two more, if they died aswell i would never, ever put more memory in before switching the board out. personally, i have almost nothing die, and i'm pretty rough with hardware(not rough but no antistatic precautions, carpeted room, not to carefu), some people might just be being completely idiotic with their ram, who knows.

in this thread which is aimed at a specific problem there are only a few people that have had an issue, and TWO of them used 3.2v on the 5v line, something that IIRC angry or windwithme reported as an officially stupid idea a while back in a thread just to tell us that. using anything below 3.3v is very bad, its the HARDEST setting for the motherboard to use, 4v is FAR LESS DANGEROUS for the board itself than 3.2v when using the 5v jumper.

i dunno, maybe there is an issue with the board. remember, you must think more people, on average that buy this boar,d say 50% of all people getting this board are running huge overclocks increased voltage on everything, where as say maybe 20% of asus sli users are using really high voltages, and more people are buying the dfi board, to see more issues is expected, completely, IMHO.

Some of the things i've seen friends do makes me laugh and cry at the same time, they can kill hardware with pure crazyness, someone dropped water into his case(glass over edge of table) and didn't htink to turn off computer straight away as he was mid game. once water hit the wrong part, died. shock horror. I'm just not seeing this huge number of issues that are being talked about.

i've got two boards, both at high clocks using utt, one i play about with gskill and the 3.3v jumper too. its a 24/7 rig the first is the sli version which i've had since the day it was available in the uk. its been off for about 2 days when i moved the system to uni(moved in mid year), and not an issue. i've got the ultra-d at home and had that for about 2 months, 24/7 as i leave it downloading while i'm away(uni with most ports blocked so still downloading at home).

PS, i'm making a huge assumption, but if both you guys sent the boards to me for rma'ing and told me you used 5 v jumper and 3.2v setting, i could straight up tell you it was stupid idea and ask why you would do it considering you could do that on the 3.3v rail with no risk at all, or i could give you any old line so i don't have to be rude on the phone and risk losing your business, i know which way i went more often in rma department. if you both got replacement boards even though you both went against recommended use then IMHO demanding another "fixed" board with zero proof of any problem is ridiculous, i'd be chuffed that i got a replacement in your situation.

was gone out for a bit. My motherboard is fine. It's my RAm that died. Yes, using the 5volt feed and only using 3.2 volts is stupid, but I never knew up until 2 weeks ago that it was actually a 5v feed. I saw it as a way of unlocking the vdimm options up to 4v, and never comprehended that it was actually using a different, and possibly less stable, power source.
05-27-2005, 09:37 PM
mcbarnet007

Are you guys who's using 5V option cooling the heatsink on the upper right corner? It gets pretty damn hot so that's prob causing the trouble. Another thing is that some Winbond UTT is very easy to kill without proper cooling at 3.3v+. I mean, come on, they are untested. I have huge copper heatsinks on mine and a 90mm fan blowing at them to keep them cool and safe. I've been using 5v option since day one and zero prob so far.
05-27-2005, 09:41 PM
mcbarnet007

Oh, why don't you have the bios limit voltage from 3.3v to 4v when jumper set to 5v? that way user will be forced to use 3.3v and not 3.2v
05-28-2005, 12:00 AM
topboy

To all

Anyone who is having problems with the 5volt jumper could you please also post over at this thread please

http://www.dfi-street.com/forum/showthread.php?t=12342

This may make the "Tech Support wake up

Thanks

Topboy
05-28-2005, 12:30 AM
felinusz

I'm of the opinion that we need to find out how many of the users who experienced problems (dead RAM) ran their boards for 24/7 use with the 5V jumper, and how many users who experienced problems only ran their boards periodically, for benching sessions, or minimal daily useage.

This information will almost completely clear up the "it's the end user's fault" catch-all.

I strongly suspect that it is mostly people running the 5V jumper for 24/7 use who are getting dead RAM.

Someone who benches for two hours a day, making the statement "I've been on the 5V jumper since day one without issue" does not mean as much as when a 24/7 user says the same thing - it actually even skues the example base that we have (I'm not bashing anyone :)).

DFI is a great company - but even great companies sometimes have product issues. I've had problems with DFI boards before (Went through two Infinity NF2s, one of which never even posted), namely the shoddy Vcore regulator used on their NF2 boards, that was only rated for 1.8V. That really pissed me off, and was inexcuseably cheap. They're still the best enthusiast motherboard manufacturer out there, hands down.

My point being - run your board with the 3.3V jumper tweak, overvolt your +3.3VDC VI/O rail to 3.8V, and do your VDIMM overvolts oldschool untill this is resolved. Enjoy your DFI boards, because there is NO other nForce4 motherboard as enthusiast oriented and thorough as this one is. IF you're not willing to run a +3.3VDC VI/O overvolt to get your high VDIMM, you shouldn't be running a high VDIMM in the first place :).

My $0.09.
05-28-2005, 12:35 AM
felinusz

I'd also like to add that launching a crusade on the DFI-Street forums is counter-productive at this point.

DFI reps have said several times that the issue is being seriously looked into. They have said several times that they will not deal in speculation, which is highly commendable and professional of them in this situation.

They are handling this as they should be - just look at some of the reactions in this thread and others. The second a DFI rep says something, it will get smeared all over the internet.
05-28-2005, 12:55 AM
G H Z
Quote:

Originally Posted by bigtoe

Hipro

We have tested the board with the booster also and it does clock better. Interesting to see your booster has some good results.

Maybe this could be the answer, use a booster/maximiser and leave the jumper on 3.3V line?

Couple of problems with running the booster on this board.
- 24 pin PSU
- VDIM readout does not work on the booster running 20-24 pin adapter.
Maybe someone needs to try a vio mod on this board now if jumper is in fact a problem.
05-28-2005, 01:08 AM
felinusz
Quote:
Originally Posted by G H Z

Couple of problems with running the booster on this board.

24 pin PSU
VDIM readout does not work on the booster running 20-24 pin adapter.

Maybe someone needs to try a vio mod on this board now if jumper is in fact a problem.
- With an OCZ PSU, the extra four pins can still be directly plugged into the motherboard because they're seperate to begin with... another reason to use an OCZ PSU I suppose.
- No need for an adaptor with an OCZ PSU ;). Alternatively, modifiying your PSU for a seperate plug isn't that hard if you don't care about warranties.
You raise an excellent point about VI/O mods. Variable output PSUs are extremely common now, almost standard in fact.

With a simple VI/O(+3.3VDC)=VDIMM voltage modification (connect the +3.3VDC pin on the PSU header block at the rear of the motherboard to an appropriate VDIMM point), VDIMM would still be completely adjustable, completely independant of the BIOS, just via direct manipulation of the PSU's +3.3VDC rail through the built-in external rail adjustments. You want 3.6V of VDIMM, you set your +3.3VDC rail to 3.6V... and I/O components can typically handle the rail overvolt within reason :).

Might solve the cold boot issue as well, and would definitely, 100% fix the 5V jumper problems. VI/O=VDIMM mods pretty much render a motherboard's VDIMM voltage supply circuitry "harmless", as it is "unused" after such a mod.

Food for thought.
05-28-2005, 01:19 AM
tictac

AG just replied.... :)

http://www.dfi-street.com/forum/show...6&postcount=85

Quote:

Memory Death:

We use two stages of linear regulators to generate the VDIMM on the DFI NF4 motherboard.

1st regulator from the psu’s 3.3v or 5v line to VDIMM sets the target voltage to 4.1v, or 3.2v (depending on what how the J17 jumper is configured). The 2nd regulator controls VDIMM voltage to the memory modules themselves.

Explanation as follows:

In 3.3v source jumper mode, the 1st regulator will output 3.3v to the 2nd regulator. The 2nd regulator handles the voltage output to the memory modules. (2.5v-3.2v)

In 5v source jumper mode, the 1st regulator will output 4.1v to the 2nd regulator. Then the 2nd regulator will handle the output to the memory modules. (2.5v-4.0v).

ATTENTION:
The DRAM voltage as originally supported by the spec is up to 2.85v only. Jumper J17 allows you to select beyond the defined spec. Although the function is supported (at the hardware level), selecting a higher voltage may cause unstable current supplied to the system board which may subsequently cause damage to the CPU and DIMMs

(copied from the warning sheet supplied with the DFI NF4 motherboard concerning the J17 5v jumper and also found on page 9 of later revision NF4 motherboards (A02 and later))

Using this design, the 5v source voltage will not exceed the VDIMM source from the 1st regulator (4.1v). This circuit will not supply DRAM voltage higher than 4.0V(4.1V-0.1~0.15V) no matter what kind of user error or bios error.

Using this type of linear voltage regulation is very resistant to any voltage spiking and in the DFI labs and in independent testing no spiking at the memory sockets has been observed.

To conclude, there has been no evidence that the DFI NF4 motherboard has been responsible for any memory death. You can choose to not believe this if that is what you want to do. We have done extensive testing and found no evidence that the board kills memory modules though, and that is the truth, no matter how much some of you might want to flame and shout that you 'know for sure' that the board does kill memory.
05-28-2005, 08:08 AM
jiff

Only problem I see is IF its stated as a DFI NF4 problem, how come I know of three incidents of DFI NF3 users with cold boot/dieing sticks.
My finding...... starts as a cold boot problem then ends up a dead stick.

My :2cents: I think this will only be resolved when the maximers comes out and sticks continue to die, then at least it will rule out the M/B once and for all
05-28-2005, 08:10 AM
pershoot

oskar wu's reply:
http://www.dfi-street.com/forum/show...9&postcount=90
05-28-2005, 08:37 AM
The Mofo

Ive not been affected by this problem but i know a few people here who have. I just hope that DFI figures out how to stop this problem so we dont end up with the only GREAT motherboard is existance. I read Oskar's post and it appears he has found the problem. I just hope the future DFI boards dont kill the +5V option.

Show 100 post(s) from this thread on one page