DFI admits 5V vDIMM problem

ah....the reasons that i am trying desperately to avoid a DFI board in my next system.......
 
I don't see why this is a reason to avoid a DFI board, as this is a problem with an optional tweak, and I am not sure what other boards sport a 3.3+ vdimm option (sure beats a really expensive booster). Someone should still be able to squeeze a good bit out of winbond memories with just 3.2 volts. I must admit though, I am going to purchase a nf4ut in the next day or so, and I for one won't set a 5v jumper without some more serious reading.
 
that is far from the primary reason that i am not planning on getting a DFI board.

as you might guess from my handle, i currently use a DFI board. my two key issues with it are 1) the build quality if very poor, demonstrating lousy manufacturing and QC practices and 2) it won't OC at all.

i know very well that my results are atypical, but i don't buy a second time in a row from a company that has screwed me.

right now i am looking at the BFG board because it is cheap, OCs like a champ and has all of the features that i am looking for (i don't care about SLI support). even though it does not allow for extreme over-volting on the memory, i use TCCD based sticks which dontt benifit from high voltages to begin with.
 
So let me get this straight...the problem is not with the board's functionality, but rather with an inherent danger associated with an end user purposefully setting their RAM voltage to 5V?

Isn't the specified voltage for DDR 2.6V? :confused:
 
changing the ram voltage should not cause the chipset to over-volt and die.

i have little sympathy for those who set their ram voltage really high and end up with a nice set of paperweights as a result, but this is not something that should be happening as a result of over-volting the ram.
 
How is it not a result of overvolting the ram? :confused:

The user decides to increase his RAM voltage from 2.6V to 5.0. This, to me, implies "overvolting" his RAM. And not just a little bit...but a real shitload! That's crazy...5.0Vdimm?? I think my old 486DX2 used that much for its RAM...but that was when transistors were the size of paper clips!
 
SKYY said:
So let me get this straight...the problem is not with the board's functionality, but rather with an inherent danger associated with an end user purposefully setting their RAM voltage to 5V?

Isn't the specified voltage for DDR 2.6V? :confused:
Click to expand...
Yes it is 2.6 with JDEC boards, most winbond modules paired with brain boards respond very well to high VDIMMs (and don't fry most importantly, I'd imagine 3.3v would kill a good number of memories). I'm pretty sure this is how it works, the jumper doesn't give the RAM 5v, it runs it of the 5v rail (don't think it'd be easy running 3.3+ volts on the 3.3v rail) enabling vdimms above 3.2
I don't believe the problem is the high Vdimm, but the way the board is supplying it. After some reading this also may be related to a cold boot problem, and of course the board/memory part failures. I also read there was a different trick to get 3.3+ volts that didn't involve the same risks, though I'll have to check up on it, as it'd be nice to have the option and gain an extra 10-20mhz, but not at the risks I've seen so far with the 5v jumper.
 
SKYY said:
How is it not a result of overvolting the ram? :confused:

The user decides to increase his RAM voltage from 2.6V to 5.0. This, to me, implies "overvolting" his RAM. And not just a little bit...but a real shitload! That's crazy...5.0Vdimm?? I think my old 486DX2 used that much for its RAM...but that was when transistors were the size of paper clips!
Click to expand...
the problem is that it IS a result of over-volting the ram.

over-volting the ram is causing the chipset to blow.

the ram and the chipset voltage are supposed to be seperate and the fact that changing one is having a dramatic impact on the other is a definate problem, not entirely the fault of the user.
 
At least they found the problem and it will be fixed in the next mobo. Which is good.

I will deffinalty be getting a DFI mobo for my Dual core CPU. I need to voltages
 
USMC2Hard4U said:
At least they found the problem and it will be fixed in the next mobo. Which is good.

I will deffinalty be getting a DFI mobo for my Dual core CPU. I need to voltages
Click to expand...
As far as I've read, the only admission of a problem was over the phone, and no fix has been presented yet. Hopefully it's the sort of thing they can fix with a board revision (current owners would still be SOL), but if it's fundemental design issue, the fix could be a generation out.
 
so what exactly is the problem? is the cpu gettin fried or something? i have the jumper setting to allow 5 volts so i can overclock my ocz value vx as it needs 3.4-3.6 volts to do well.
 
It happens that with my current mem/cpu combo 3.2v is enough to get to 2800MHz and no amount of voltage gets me stable much beyond that so I can live with 3.2. I'm going to play dumb and not waste energy worrying over problems that have small impact on my current situation. That said I noticed zero voltage fluctuation while I ran 3.2+ and I monitored and charted it for a number of hours.You'd really need to scope it and I don't know what the sampling interval is for these monitoring programs so take it all with a grain of salt.
 
you're only getting 215 mhz @ 2-2-2 with your UTT twinmos? I thought most hit 240+?
 
I think some of you guys are misunderstanding the problem. No on is setting there vdimm to 5.0V. That woudl be straight up insanity and guarenteed to fry alot of stuff. This board does not allow for anything over 4v vdimm. What it is, is the jumper on the board that allows you to have the ram powered from the 3.3v rail or the 5v rail. on the default settinngs, the ram is suplied with power from the 3.3v rail, and you can't set vdimm higher than 3.2v. Change the jumper and the ram with be supplied power from the 5v rail, and you can select a vdimm up to 4v. Changing the jumer to get power from the 5v rail is the problem, NOT people setting their vdimm to 5.0v, and thus no one is being wreckless as some of you have implied. Memory such as UTT and BH-5 needs 3.4-3.6v to reach its prime, and with propper cooling that is not being out of line, but it would require the 5v jumper to be set to get those voltages, which could kill the board without the user doing anything wrong
 
mikelz85 said:
you're only getting 215 mhz @ 2-2-2 with your UTT twinmos? I thought most hit 240+?
Click to expand...

I have a thread on this (fine tuning San Diego) I can get up to ~250 with 3.5V but I have to lower my cpu divider and lose 50 -100 mhz or so on CPU. System doesn't like high mem and cpu at same time. Best I can manage is 3.5V 9/10 divider on mem and 1.55V 10.5 on CPU x 267mhz htt. The cpu ~2800 mhz on both but the latency drops from from 4 to 3 on the internal cache with the higher MHz.

For now I'll stay at lower voltages and temps and lose a bit off the memory.
 
Viper87227 said:
Memory such as UTT and BH-5 needs 3.4-3.6v to reach its prime, and with propper cooling that is not being out of line, but it would require the 5v jumper to be set to get those voltages, which could kill the board without the user doing anything wrong
Click to expand...

so basically, setting the board to allow up to 4v for memory and spontanously kill the board and maybe the cpu as well?
 
peterhoang2002 said:
so basically, setting the board to allow up to 4v for memory and spontanously kill the board and maybe the cpu as well?
Click to expand...

Yep. For most people its fine, but some people have problems. I had it happen to me. Its doesnt always totally kill the board, but it can screw it up pretty bad. for example...when I first got the DFi board, I had smoe Bh5 to run in it. At 3.2v, i could run up to 240mhz 2-2-2-5. I then changed to the 5v setting, being careful to follow the instructions, and after I did it the system came on and then back off....it was like it was being shorted. Trieda few more times, same. at this point i was scared shitless coz i just dropped $1500 on this puter, so I set the jumper back to 3.3v, and it came on but wouldnt post. I reset the bios and it worked fine, but from then on even with 3.2v, my ram never budged above 210mhz. Tried the memory in another system and it worked great, tried a different set of bh-5 and a set of tccd in this board and got nothin gbut problems. Since then, i even got a new CPU and still nothing but problems. Right now I am using the chaintech board in my sig...i just need to get around to RMAing this board.

EDIT: Just thought I woudl clarify, if you want more info on this, good the phrase "DFI" + "Sudden Death". You shoudl find pletly of people who had this problem.
 
This is disappointing as my OCZ VX Gold needs 3.4-3.6 in order to max itself out at 240mhz 2-2-2-8 or better - and it also needs the extra juice to keep up with the memory controller in my venice at 280+ HTT.

Oh well, back to 3.2-3.3v and I'll just have to find the optimum overclock at those juice ratings.
 
For what it's worth, it looks like the official DFI support forums are saying "there's no proof that the 5v kills ram or chipset".

So it's really just speculation on behalf of random internet people, and some a DFI phone support rep (I do call center work myself, and readily admit the info I get may be incorrect or personally biased - or phrased in such a way that's not totally incorrect and simply just a lack of clarification).

So take it with a grain of salt until there's an official DFI stance. There's a LOT of people running 3.4v or higher that haven't had any issues.
 
its funny how only a handful of people have this problem and all of a sudden its everybodys board.

and if you been following this for a while its was at tech over the phone that said this not a pr person from DFI but a tech.
and that guys story has changed a few times since he posted that.
 
Filter said:
its funny how only a handful of people have this problem and all of a sudden its everybodys board.

and if you been following this for a while its was at tech over the phone that said this not a pr person from DFI but a tech.
and that guys story has changed a few times since he posted that.
Click to expand...

What's to say that a tech is absolutely 100% correct, never wrong, and totally fool-proof?

I'd hardly say RGone or Angry_Games @ DFI-Street.com are PR and not tech. Quite the opposite infact.
 
Lukano said:
What's to say that a tech is absolutely 100% correct, never wrong, and totally fool-proof?

I'd hardly say RGone or Angry_Games @ DFI-Street.com are PR and not tech. Quite the opposite infact.
Click to expand...

and thats why many threads on this subject have been deleted or locked over there??
 
Filter said:
and thats why many threads on this subject have been deleted or locked over there??
Click to expand...

What relevance does that have in regards to whether they're PR and not Tech?

Tech's get fed up when the same thread is posted time and time and time and time again when the information is readily available (sticky, other posts on the front page of the forum, or easily found with search). Same goes for unfounded posts with no confirming documentation or evidence.

No need to go all tinfoil hat simply because threads are locked, moved, or deleted. It's not a conspiracy theory, Really!
 
Well, boards aren't dying from nothing, and it'd be nice to hear DFI acknowledge the problem/issue but it's very unlikely. After reading in my opinion there is more than some risk associated with the 5v jumper. I completely understand that there are risks with running memory at 3+ volts, but frying chipsets is something completely different.
 
sounds like DFI's theory on the dead memory is centered around migration, which doesn't suprise me given how much juice is being pumped to these memories in some cases. When I first saw and researched the explosion of winbond memories I wondered how companies such as OCZ would insulate themselves from the risk of huge RMA numbers down the road from such high vdimms, as they do spec the ram to run the high vdimms, and even sell voltage boosters to run the voltages if boards don't natively support them. I mean how long with this stuff last at the highest support voltages, especially without active cooling? 6 months? 12 months? Either companies fully expect these memories to keep working and are correct, or are wrong, or expect some of these to fail and be RMA'd them and build this into the original price of the item.

On another note I think this thread title is a little deceptive given what DFI is saying in this thread here: http://www.dfi-street.com/forum/showthread.php?t=12687&page=1&pp=15
or at least contradictory
 
mikelz85 said:
http://www.dfi-street.com/forum/showthread.php?t=12687&page=1&pp=15
or at least contradictory
Click to expand...

That was my thought too.

I think there's a lot of tinfoil-hattery and speculation going on here that has no purpose other than provide an avenue for those who dislike DFI to vent, and those who like to stir the pot to - well, stir the pot. This thread, as well as threads on other forums all seem to be the same thing. Little constructive information, advice, accounts, etc - and more beaking and griping and sniping at DFI for (what would be, even if it WERE to end up being true) a common and industry acceptable level of a mistake. Other companies have made worse mistake, and more often. And that's making a huge assumption that all this hulabaloo is correct and not a gigantic game of telephone resulting in misinformation.
 
^^^ i agree.

also, i'm fairly certain that dfi advises against using the 5v jumper for 24/7. it's there mostly for benching purposes. if you really want a stable ram voltage over 3.3v, do what us s754 dfi users do and crack the 3.3v rail. i've never heard of anyone killing something by doing that yet ;)


ALSO!
why buy a dfi and overvolt your stuff if you're afraid of killing your hardware and will point fingers when something fails instead of admitting that it's mostly your fault? :p
 
(cf)Eclipse said:
ALSO!
why buy a dfi and overvolt your stuff if you're afraid of killing your hardware and will point fingers when something fails instead of admitting that it's mostly your fault? :p
Click to expand...


QFTT
 
It sucks that a nice feature doesn't work out so well, but that sometimes happens when you're doing new things.

Anyways, I was going to post this before, but wanted to stay out of the madhouse. Andy at OCZ posted up a method for getting past the 4Vdimm jumper, and it's been stickied for a while at the Street. I know that modding power supplies is common knowledge to a lot of overclockers, but here it is anyways... http://www.dfi-street.com/forum/showthread.php?t=12085
 
Has anyone here thought about the fact that maybe it's not the motherboard that has the problems but instead it's the memory? As far as I see it, it's not the motherboard killing the memory it's the high voltage the motherboard allows users to supply to their memory.

The winbond chips the dead memory uses is not designed to run at such high voltages. According to the specifications the maximum voltage of the winbond chips is 3.6v. Most of the memory manufacturers only warranty the memory to a max of 3.5v and the specs of the memory calls for 3.2v.

The fact the DFI allows users to pump 4.1v into there memory defintiley does not mean its recommended or supported by the memory. In fact that much voltage is not at all recommended by the memory manufacturers. Until someone can prove that it is the motherboard itself causing the destroyed memory then a reasonable assumption is the memory simply can't withstand the increased voltage that it wasn't designed to handle. Until there is proof that the motherboards themself are causing the dead modules and not the voltage then this problem isn't DFIs fault.
 
darn_it said:
Has anyone here thought about the fact that maybe it's not the motherboard that has the problems but instead it's the memory? As far as I see it, it's not the motherboard killing the memory it's the high voltage the motherboard allows users to supply to their memory.

The winbond chips the dead memory uses is not designed to run at such high voltages. According to the specifications the maximum voltage of the winbond chips is 3.6v. Most of the memory manufacturers only warranty the memory to a max of 3.5v and the specs of the memory calls for 3.2v.

The fact the DFI allows users to pump 4.1v into there memory defintiley does not mean its recommended or supported by the memory. In fact that much voltage is not at all recommended by the memory manufacturers. Until someone can prove that it is the motherboard itself causing the destroyed memory then a reasonable assumption is the memory simply can't withstand the increased voltage that it wasn't designed to handle. Until there is proof that the motherboards themself are causing the dead modules and not the voltage then this problem isn't DFIs fault.
Click to expand...

I think in the very FEW cases where there has been damage, it hasn't been the motherboard or the memory that have been damaged - but the chipset. More than likely due to poor airflow or passive/active cooling and subsequent damage from overvolting.

I really can't see anyone pumping 4v into their memory for any purpose, and if they do they damned well deserve what they get. :)
 
i wanna know what exactly these people that are having problems were doing

how high did they overvolt?
were they coling the ram?
did they have a quality psu?
hows the rest of the case's airflow?


there seem to be lots of opportunities for confounding here, and it seems a bit premature at best and shortsighted at worst to blame DFI right away
 
You must log in or register to reply here.
Back
Top