DFI NF4 the UTT/VX/BH5 KILLER READ IF YOU HAVE DFI

[hipro5]

Most of the DDR SDRAM only gurantee 2.8V maximum absolutely rating working voltage ... Anything beyond that is an unknown status ...

Winbond BH-6 , BH-5 , CH-5 , UTT........3.6Volts max......

[hipro5]

well, with all the knowledge that i have, and some other guys that are specialist in hardware, that includes ram memory.. i can pressume that if you want a high overclock for 24/7 use, you can't never reach your mems to more than 3.0 volts, cause even at 2.9v the ram its hitting almost the 50 centrigrades wall, that for what i know its like the temperature risk start that can cause fails and deaths on your sistem components, you named, cpu, chipset,ram.. etc

so getting a samsung tccd ram chipset..can give you an edge of a high overclock with low voltages, so i guesss that these ones are the real kings of overclock , cause al winbond chipset are made for overclock benchmarks and that's all..

WRONG.........I'm with 2 x Winbond BH-6 @ 3.45Volts 24/7 crunching first SETI and now D20L......NO problem at all.......Also SETI is working Rams a LOT and it is VERY memory depentant......
On my other system also BH-5 and BH-6 (one and one) is crunching 24/7 @ 3.35Vdimm.....

EDIT : I'm almost OVER 2 years with those Rams and I never have worked them BELOW 3.3Vdimm.....NEVER.....

[P_1]

hey hipro, kinda off topic but hows ur maximizer comming along? Also do u know what the cause of the deaths are? It seems to be happening to both tccd and winbond memory on the 5v rail setting and with it off so it seems like a lot of people will be affected by this.

[hipro5]

hey hipro, kinda off topic but hows ur maximizer comming along? Also do u know what the cause of the deaths are? It seems to be happening to both tccd and winbond memory on the 5v rail setting and with it off so it seems like a lot of people will be affected by this.

Off topic...it's coming to solves those problems any time soon......We had to fabricate it and assemble it in Taiwan coz the cost here in Hellas was TOO HIGH......

I think that those problems are focusing on the VERY start up of the mobo....It might "give" to Rams a big and high pulse(Vdimm) spice or so.....Next week I'll get once again a mobo my shelf so as to investigate this.....

[hovo73]

I wish we could get an acknowledgement from DFI that they are at least looking into this problem. May be we should start a petition or something? Because once, again, it is a good board, I would like to keep it, but I don't want to keep killig my RAM. I have just got my brand spanking new TwinMOS BH-5 and OCZ EL BH-5, but I am not putting them into this board which kind of sucks. I also suspect that my Patriot XBLK was also damaged damaged.

[bachus_anonym]

After reading threads of this sort, I very honestly think that DFI NF4 and OCing voltage hungry BH-6/BH-5/UTT is just not the game everybody should play.
DFI might have done this one mistake of putting board like that on the market, so EVERYONE can have it. Before, volt mods needed to be done to boards, which made hardcore/high voltage OCing not accessible to all. Now, DFI might have a lot of trouble, because of folks popping up with all sort of hardware deaths that they attribute to boards and not errors they make.

DFI NF4 is a very specific mainboard. It needs special care and fine tunning, to get it work as it's supposed to. Switching between vdimm supply is one of the crucial points, where users need to be careful.

Also, I might be wrong here, but I start to think that Winbond UTT based sticks are a gamble and can attribute to those failures that some of you might be experiencing. As I had very similar symptoms with OCZ VX PC3200 months ago on MSI Neo2. I ended up going thru 2 RMAs and at the end decided to give up on UTT. Don't you guys think that it's pretty weird that vast majority of having those problems use UTT-based sticks? Apart from eclypse, in which case I honestly think that feeding TCCD 3.2v might not be such a great idea for everyday use. Maybe he was just a guinea pig, testing limits of TCCD and found it .

I'm running my 2x256MB Mushkin LvL II on a daily basis at high volts of 3.7v, 5v supply. Benching at up to 3.9v with no problems whatsoever.

[hovo73]

bachus_anonym, don't you think that if it is true that the voltage spikes to 5V during startup it really has nothign to do with user error, and it is a valid complaint? Besides, there a lot of people with this problems and not all of them are noobs.

[bachus_anonym]

bachus_anonym, don't you think that if it is true that the voltage spikes to 5V during startup it really has nothign to do with user error, and it is a valid complaint? Besides, there a lot of people with this problems and not all of them are noobs.

You do not know if that is taking place at all. Without at least a multimeter (which many of DFI users do not seem to own) or better a scope there is no way of investigating that.
I highly doubt that this is the reason (voltage spikes) is the reason behind all the trouble you guys are having.

@George (Hipro5)

It would be great if you could take a look at it in your test-lab with a scope and see if that is the case. That could clear things up a bit

[hipro5]

One think to add too though is that I've seen some dead Rams BUT their chips where OK......Their SPD chips were dead......After changing the SPD chip with another one, Ram was working perfect again.......Maybe the SPD chip can't handle the spikes or so and it's just give up

[hipro5]

You do not know if that is taking place at all. Without at least a multimeter (which many of DFI users do not seem to own) or better a scope there is no way of investigating that.
I highly doubt that this is the reason (voltage spikes) is the reason behind all the trouble you guys are having.

@George (Hipro5)

It would be great if you could take a look at it in your test-lab with a scope and see if that is the case. That could clear things up a bit

I'll talk to esdee to bring me his mobo on Monday morning so as to see what's going on......I'll keep you posted.....

[hovo73]

You do not know if that is taking place at all. Without at least a multimeter (which many of DFI users do not seem to own) or better a scope there is no way of investigating that.
I highly doubt that this is the reason (voltage spikes) is the reason behind all the trouble you guys are having.

@George (Hipro5)

True, I haven't done it but one of the other members here claimed that he has and he has seen it spike. Given the number of people that are coming up with this problem, don't you think that it at least deserves looking into by DFI? And IF it turns out true, don't you agree that it has really nothing to do with user error?

It would be great if you could take a look at it in your test-lab with a scope and see if that is the case. That could clear things up a bit

Yeah that would be awesome if some one with knowledge and equipment could do this.

[bachus_anonym]

What's important is that when switching from 3.3v to 5v or the other way around, user should LOAD DEFAULTS first, so vdimm goes down to say 2.8-2.9v. Then go to BIOS and shut it down there, drain caps with POWER BUTTON. Then Clear CMOS, unplug everything, including power plug, switch the jumper.

@hovo73

I've already asked that guy where he saw that 5.2v spike but did not get any answer so far in this thread.

EDIT: Also, if one is REALLY paranoic about it, he can even first start-up after switching over, boot up without any memory sticks. Then shut it down, at put sticks in

[hovo73]

According to DFI-Street when changing J17 position you should do the following:

Changing from 3.3V to 5V:
http://www.dfi-street.com/forum/show...7&postcount=17

1. Go into Genie BIOS and manually set your vcore (cpu voltage). It MUST NOT be on Auto!
2. Save & exit. Return to bios upon reboot.
3. Verify that your vcore is correctly set. Save, exit, shut the rig down.
4. On the motherboard, move the 4v jumper in the upper corner of the board to the proper position (see manual/cd-manual/online manual for pics).
5. Boot up, go into Genie BIOS, do DRAM Voltage. Should now be

Changing from 5V to 3.3V:
http://www.dfi-street.com/forum/show...05&postcount=4

1. boot into bios and lower vdimm to below 3.3v
2. save & exit
3. boot up and see that the voltage 'stuck' to the new setting correctly.
4. save & exit and shut down
5. change vdimm jumper back to 3.2v position.
6. boot up and make sure vdimm is where it is supposed to be in the bios.
7. keep in mind that sometimes the voltages for vdimm or cpu 'stick'. Its very random and rare but I have witnessed it myself. Its a quirk. Clear CMOS and the voltages will return to their defaults and can be reset.

Then go to BIOS and shut it down there...

You can shutdown the PC when you are in BIOS? That is something that interested me, because it is recomended to do it aimmediately after flash and I have a thing about shuting down the PC while it is posting. Is it safe to shut down the PC (using the switch) when it is in BIOS?

[bachus_anonym]

If you flash from Windows, you can shut it down there, after Winflash asks you to REBOOT or not.

If from DOS prompt, then depending on how the awdflash is configured in .bat file (it can RESET automatically or not) you can shut it down there, after it's done flashing (if it doesnt reset itself).

I don't think there is a problem with shutting down while in BIOS.

[coop]

I agree, flashing the bios in any state other than 'optimised defaults' is just an additional risk. That includes having the jumper on the high voltage side. It also seems more of an issue with certain rams. I'm glad dfi made this board. With my fat hands I don't solder all that great.

[Nugz]

hipro5 is going to be on it. think we might have a fix comming. woot Iv seen 3-4 total MB vmods from him. if i gota cut a trace, ill do it lol.

thanks for finally lookiing into this hipro5

Peace Nugz

[celemine1Gig]

Well, that's really interesting, because I already thought about getting that DFI board some time soon, but now, I'm really not sure. Although I doubt that it really is a voltae spike that is causing this problem. First I think DFI did really thorough testing and I also think that their engineer team knows how to aviod this kind of problems from the start.
Do you remember the Shuttle AN50R and all the dead CPUs it caused when VDimm was modded? Well, on this board, there were voltage spikes for sure, but they only killed the CPUs, but I can't remember having heard of dead RAM. That's what makes me wonder.
Hipro's theory of SPD-ROM chips dying sounds more reasonable to me. I had a look at my UTT sticks today and found out that the SPD-ROM chip used on mine ("TMC 24A02") is rated 2.5~5.5V. So on mine, there should theoretically be no problem with high Vdimm.

Just for fun:

Could anyone with "dead" RAM post the info printed on his SPD-ROM chip? Perhaps this way, we could find out quite fast, if the theory is any good.

[jinu117]

What's important is that when switching from 3.3v to 5v or the other way around, user should LOAD DEFAULTS first, so vdimm goes down to say 2.8-2.9v. Then go to BIOS and shut it down there, drain caps with POWER BUTTON. Then Clear CMOS, unplug everything, including power plug, switch the jumper.

@hovo73

I've already asked that guy where he saw that 5.2v spike but did not get any answer so far in this thread.

EDIT: Also, if one is REALLY paranoic about it, he can even first start-up after switching over, boot up without any memory sticks. Then shut it down, at put sticks in

Bachus, your eloquent comment about NF-4 needing upmost care, I partially agree. However, you keep skipping over the problem of memory dying NOT DUE TO BIOS FLASH, but just act of booting issue from cold start after about 3 weeks of 24x21 use with case not even bothered to be opened (no bios change, no jumper change or what not) and to find out your memory is dead, doesn't fall under what you are referring to. In fact more than half of us who has complained the issue didn't change anything from fine working daily setup we aspire to other than just starting the machine from cold. Now, if you don't call that an issue, oh well....
PS) I know this is getting personal but we had similar kind of discussion before on DFI nf4 issue Bachus, when I didn't get working product first shot at all. YOU BLAMED my ignorance and my ineptitude... (without knowing even much of me at all). And yes the problem is well reknown (now) 3 led syndrome. Just because yours works and works well doesn't mean others with similar experience level (or even more) than you on building and overclocking system SHOULD have no issue. This holier than thou attitude is really starting to get on me.

[jinu117]

It seems that UTT(UnTesTed) from winbond is not able to handle extremely high voltage as old-good BH5 or CH5 ... It seems that UTT is not 100% the same as BH5 because BH5 can not boot in CAS=3 setting , but UTT can boot in CAS=3 setting , maybe there is something different from UTT and BH5 to cause that UTT not able to handle extremely high voltage

I should have elaborated. I had my machine running for 3 weeks straight with 3.4 or 3.5v VDIMM and when I had to reboot for windows update, I thought clean cold reboot would be nice after such long marathon session on my comp. It just didn't boot up after that. From perfectly working sticks to 4 dead sticks (yes all 4) in matter of few seconds it took to boot from cold? ALL 4 STICKS dying together? Doubt it is my memory's getting too much juice alone... its gotta be something else. (If it was one stick or so, I would have thought it is memory but oh well) Now considering how people sticking to 3.2v or less without 5v jumper on, doesn't have this issue, it becomes quite obvious to me that this is with 5v jumper issue of some sort. That voltage spike idea with VSB might be something to look into, for now, I just don't know.

[bachus_anonym]

That BIOS talk was a result of me being side-tracked a bit

I will bet $5 that it's the issue with memory sticks, not mainboard itself. Because, as I said above, I had two kits of OCZ VX sticks that died same way, but on the Neo2 not DFI. One set I got, was good for about 2 weeks, then I stick got fubared. I checked it on Neo2 and NF2... One stick was memtesting fine when the other was giving flood of errors all the way down to stock. Second kit I got, was same story... Good for three weeks, died same way...

I really think that UTT mems is what causing the trouble. As most of you have that problem with that type om sticks.

EDIT: I admit, this i pretty weird though...

[pazza316]

Guys my BH5 ran 280 FSB 2-2-2-5 for a year plus at 3.7 / 3.8 in both DFI UT250 and MSI NEO 2. I also ran one set of my VX at 3.6 2-2-2-5 in the same mob's at 260 FSB with 3.6 volts. Put them in the DFI 5volt rail jumper setting and both sets Dead within 3 weeks. The board is at fault not the memory this BH5 would run stable at 4 volts!! and 3.8 was 24/7!!

[pazza316]

I will post the codes on my SPD chips when I get home from work Sunday. Then that theory can be checked.

[eclypse]

Damn i used 3.24v on my GIG of G-Skill PC4400 LE TCCD ram just to get 270 HTT stable. Am i over juicing it? Sure i thought this ram would run at a lower Vdimm of 3v at a high 270+ FSB. Is this not the norm?

[Viss]

Ive been running me VX 24/7 @3,5volt at 250-260 from day 1 (januari) without any problems (cold boot etc) so far. Im using the 5v line.

[Millyons]

OK i know i already posted this thaugh in another thread, but why did DFI even implement the 3.3 / 5 switch why not just run the vdimm of the 5v like some epoxes (im not a epox fanboy)

if it was for the inexperianced, they coould have hidden the high vdimm options through bios or just make us flash modded bioses with unhidden features..........