DFI SLI DR Expert CPU Death Fix

I do not think that the Expert has the battery problem that was troubling the RDX - if I remember correctly I tested the Expert quickly for that bug and could not find it. But then I was running a later beta bios - which potentially was hiding that bugger.

BUT as long as DFI doesn't officially allow to remove it - it would best to avoid it - to be on the safe side...

The most worrying bug on the Expert to me was showing itself the strongest was the returning from sleep mode temp jump - which is quite extreme. Talking 15-20Cs. As a guess instead drawing power form 8 phases it prolly only gets it from less like half - something very funky happening. Maybe something to do with the split rails? When this or other situations will trigger for that to happen I could see big voltage spikes - that make PUFFF to fragile cpus.

This was all just a guess

- done with looking into DFI boards - was actually evaluating both the Expert + RDX for implementation in a OEM project - cost us some - but DFI a LOT! At least 500 board order from their Industrial board division alone...
Now I have a few of boards that will go on ebay now...

But on a lighter note - both boards have a lot of potential once fully debugged.

Quote:

---

The most worrying bug on the Expert to me was showing itself the strongest was the returning from sleep mode temp jump - which is quite extreme. Talking 15-20Cs. As a guess instead drawing power form 8 phases it prolly only gets it from less like half - something very funky happening. Maybe something to do with the split rails? When this or other situations will trigger for that to happen I could see big voltage spikes - that make PUFFF to fragile cpus.

---

If only half the power regulation is comming online on a cold post or from sleep that would definatly cause some sparks and smoke... hm.

Quote:

---

Originally Posted by STEvil

If only half the power regulation is comming online on a cold post or from sleep that would definatly cause some sparks and smoke... hm.

---

That's precisely what my guess is. That's were the 90mn cpus will suffer, even more so when already high voltages are in place to begin with.

Quote:

---

Originally Posted by saaya

if i got it right what killed the boards, according to dfi, was that people removed the battery and did NOT set the jumper to reset. does somebody understand how this could kill the board?

---

That is the RDX board only, not the Expert. What DFI Taiwan said is that removing the battery makes the board lose the BIOS defaults. Not the CMOS settings, the BIOS defaults, so that if the boards ends up trying to start from scratch and loads from the BIOS defaults instead of the CMOS user settings, it comes up with random values.

Personally I don't believe a word of this, it is obvious rubbish.

There is no way that they built a board which wouldn't survive the battery go dead some time in the future. The BIOS defaults are obviously stored in the flash chip along with the BIOS code. It's not that DFI goes and writes their own BIOS.

Super Nadem dfi said if the battery is removed and the jumper is not set to reset but normal then the "default values" get erased, and this results in the voltages beeing way too high i think.

but where are the default values stored then? in bios?

Martin Cracauer, hmm thats basically what i think as well...

just wanted to thank all of you, mike, cracauer, stevil pcdog etc for the heads up to all the people in this thread :toast:

hope we can keep pushing this until we find the issue and dfi finally gets rid of the bugs so we can all enjoy this great board :D

Quote:

---

Originally Posted by Martin Cracauer

That is the RDX board only, not the Expert. What DFI Taiwan said is that removing the battery makes the board lose the BIOS defaults. Not the CMOS settings, the BIOS defaults, so that if the boards ends up trying to start from scratch and loads from the BIOS defaults instead of the CMOS user settings, it comes up with random values.

Personally I don't believe a word of this, it is obvious rubbish.

There is no way that they built a board which wouldn't survive the battery go dead some time in the future. The BIOS defaults are obviously stored in the flash chip along with the BIOS code. It's not that DFI goes and writes their own BIOS.

---

Well I can confirm that pullling the battery on the RDX does screw up the voltage settings. BUT the jumper has NO impact on this. At least when the sequence is like that:

1. Unplug PSU
2. Jumper to clear
3. CMos Remove battery

Wait a bit

4. put back battery
5. Cmos Jumper to save
6. plug back PSU


Again only RDX.

BUT on the Expert I caused another lil' fatality - accidently once hit both reset and start button at the same time - killed one stick of ram. Buttons are close together and can happen...'specially with big fingers...lol

Just small contribute : I never remove the battery on the Expert before my FX-57 die on it.

Altought, I seriously doubt that the issue the cause the deads of CPU's have someting to do with with battery removal as Hipro5 and others states in this thread..

I send e-mails to DFI and PM's to DFI rep here a XS offering my full cooperation to help on the investigation of the causes of dead of my golden Fx-57 on the DFI Expert : I never receive a single response:(

Quote:

---

Originally Posted by saaya

Super Nadem dfi said if the battery is removed and the jumper is not set to reset but normal then the "default values" get erased, and this results in the voltages beeing way too high i think.

---

If this is a reproduceable problem, there is more to the matter. It's not hard to probe the contents of the EEPROM to see what actually changed. Anybody know what particular EEPROMis being used?

My guess is that a variation of the INTERSIL X4C105 is being used. Most EEPROM's have built-in trip protection and AUTO store circuitry which minimizes the chance of BIOS corruption.

Quote:

---

Originally Posted by Pedro Rocha

Just small contribute : I never remove the battery on the Expert before my FX-57 die on it.

Altought, I seriously doubt that the issue the cause the deads of CPU's have someting to do with with battery removal as Hipro5 and others states in this thread..

I send e-mails to DFI and PM's to DFI rep here a XS offering my full cooperation to help on the investigation of the causes of dead of my golden Fx-57 on the DFI Expert : I never receive a single response:(

---

i have to agree pedro, the deaths do not appear related to the battery, i have tried to reproduce what mike had found on the rdx in my expert using a junky 3000 and it doesn't matter under what circumstances the battery is removed i can't kill it.

it's unfortunate that dfi would not like to investigate and examine your hardware, i would really like to see a resolution to this.

Pedro, do they have your board now?

Quote:

---

Originally Posted by loonym

It's unfortunate that dfi would not like to investigate and examine your hardware, i would really like to see a resolution to this.

---

Not only that he can't even get a response from them so far :confused:

Quote:

---

Originally Posted by Super Nade

Pedro, do they have your board now?

---

No, as I mention they didn't reply to my offer helping then to find the cause of the problem that leaded to the dead of the cpu..

Pedro, as a guess I'd say this could apply to your death:

Quote:

---

The most worrying bug on the Expert to me was showing itself the strongest was the returning from sleep mode temp jump - which is quite extreme. Talking 15-20Cs. As a guess instead drawing power form 8 phases it prolly only gets it from less like half - something very funky happening. Maybe something to do with the split rails? When this or other situations will trigger for that to happen I could see big voltage spikes - that make PUFFF to fragile cpus.

---

Loonym - 3000s are hard to kill - use a X2 and it'll be easier. Not sure if you followed the RDX thread - but the battery alone doesn't kill the CPU - it sets your default voltage to 1.55V no matter what you set it manually after reboot and adds whatever SPecial VID you choose on top of the 1.55V WHen you talk 90nm cpu it is a whole different ballgame...

Go ahead and set VID Control voltage to 1.0V and VID Special Control to 133% which in theory should give 1.33V - go and put some load on the bugger! GL - X2 it would be much easier to accomplish that

thigns are lookign better now. I resurrected my "chiped" naked core FX-57 and now have it under the Vapo LS. I thought it was long dead.

I still cant' do 250fsb on this board. So far the best I can get is 240fsb. Thats on any multi up to the 13 anythign above and it gets a black screen. I even tried 10x250 and it was black screen city. I guess some of my problem was the X2 4800. My Opty 148 will be here tomorrow.

Okay well I just got my Expert today.... Do I use the beta bios or is the official one out yet?

Also, when I download the latest beta BIOS, does it contain all the previous Bios Revisions? Like support for Opterons and the "fix" or whatever. Also did the newest beta bios fix the issue?

use thre beta off dfi site

Which one? The latest? Surely AMD has the all the previous bios updates included in the latest, right?

12/07 beta bios and a link is in the first post of this thread. Enjoy :)

Okay.... I guess the answer to my question is yes then since nobody wants to answer it!

I know where the bios is lol.

Quote:

---

Originally Posted by Tony

use thre beta off dfi site

---

And in doing so void your mobo warranty according to DFI :rolleyes: :fact:

Quote:

---

Originally Posted by alpha0ne

And in doing so void your mobo warranty according to DFI :rolleyes: :fact:

---

So what? Overclocking 1mhz voids your warranty. LOL...that's a silly reason not to try the beta.

Quote:

---

Originally Posted by mdzcpa

So what? Overclocking 1mhz voids your warranty. LOL...that's a silly reason not to try the beta.

---

Any hard facts to support this 1MHz :p:

Go over to DFI Street and its posted in just about every sticky.

Just about everything in your system is warranty voided upon overclocking...this includes the CPU, Video card (s), memory, etc etc. That's pretty clear cut.

As far as the board itself....the warranty isn't voided per se by running it out of spec, but any occurence of any problem is not covered whilst overclocking...so its really the same thing. If your board fails to boot because you are running 1 mhz out of spec, you are out of luck.

So the bottom line, here at XS where we tend to overclock a few things just a tad, and manage to void just about all the warrantys anyway, worrying about a beta bios warranty is silly:rolleyes:

Considering whether or not to run a beta BIOS in terms of stability is smart. Worrying about the mobo warranty from a beta bios on an an extreme overclocking forum is funny:D

Hmmmmmm, Void! What a lovely word. Heh! Anyone else have any download corrupted files using the top LAN port...the Giga-LAN? The 10/100 port works fine but when I use the Giga-LAN my downloads are corrupted everytime.

New bios released guys...
i now it was available as a beta bios but now here it is as a final one
download it here:http://us.dfi.com.tw/Support/Downloa...______________