Printable View
Cool, hit me bro!Quote:
Originally Posted by pcdoc1
@ Saaya, the board still works but I do not see a post screen when I reboot. Its a total black screen for 1min or so and then the Windows XP welcome screen pops up and windows is booted. I can hear the mobo post "beep." But I cannot get into the bios...yes, heh...I am hitting "delete" on reboot.
Well, I think it is safe to say that the problems that were once thought to be "user error" and "noobness" may not be that after all. It is really unfortunate, though, that a board with this much potential is causing this many headaches. I for one will be selling mine when I get it back from RMA. There is no way in hell that that thing is going back into my rig, even if the chances of it killing my components are slim. I suppose I am very, very lucky that I RMA'd it at the first sign if evil.
Shawn
hmmm thats very weird... id rather say its a videocard issue than board issue though from your description... or the monitor takes a long time to finally brighten up so you can see something?Quote:
Originally Posted by kraggy
hmmm doesnt sound like that though
can you check if its the videocard?
have another one around?
shawn, be a lil more patient, maybe they have a new revision of boards that fixes the issue(s?)
and im not surprised to see such an excellent board causing so many headaches, the more variables when configuring hardware the more likely the chance you run into problems... and pushing hardware to the edge has always resulted in weird and unpredictable behaviour.
but this issue is something else, it seems to have nothing to do with overclocking or tweaking but is rather a design flaw.
i dont know if this could be fixed in bios at all, i was kinda surprised to see a bios update supposed to fix the issue. well as i said, i have barely an idea about circuits and all i know is based on speculation and things i heard here and there. so maybe a bios can fix this issue after all... at least on the dfi nf4 ultrad the issue seems to only occur with very early bios versions and is fixed in the later bios files.
kraggy - ignore the pm if you want 12-07, looks like Hirotaro ya.......
I'm a dumb @$$'ed noob. Actually I'm on pain killers cause I messed up my neck shrugging 330lbs...never do that!:nono:
But the medication makes meh dopey:p: ...I had plugged the DVI cable in the wrong port! Shheeeeesh! I got into the bios after moving the cable but I'm dissappointed in my X2-4800...it did not boot at 3gig 250x12, booted at 245x12 but crashed, now its at 240x12= 2,880...that sucks!
From all the photos of the burned DFI mobos I've seen up to now and especially the VCore Mos-Fets blown mobos, I can say this: When we fire up the mobo, we provide to the CPU NOT ONLY the VCore BUT some other voltages as well.......Now.....When a short circuit takes place on our VCore Mos-Fets to GROUND, for some mS we DON'T HAVE VCore on the CPU, BUT we STILL HAVE those other voltages......Maybe because of this, the CPU becomes unstable and finally burns down.....
As I look at the photos of the burned down DFI mobos, I see that the burn is close to the Mos-Fet AND the heatsink of it....Maybe there is a LITTLE space between the hole of the Mos-fet heatsink the the + voltage and when the circuit starts to oscilate (4 phase), a spike goes through Mos-fet to ground and short circuit the output VCore for some mS OR for permanent, leaving the CPU without any VCore and with the other voltages as I said above........Just speculating though......BUT I think that there is a "shealed ground" all over the VCore Mos-Fets that is very "close" to the + voltage, if you can understand what I would like to describe.....:)
I wonder what DFI will do if this board starts killing cpus left and right because of a hardware issue...
and those of us that went to a beta bios for protection, not have no warranty, so i guess the hundards of us got ripped off?
I hope they would be willing to fix the problem if its hardware, or i could see some class action happening if it were to get that bad.
IF it's a hardware problem then DFI will have: a. to admit it and b. to take back ALL the mobos they have built so far so as to give a new revision to public......NOW if the above happens, we have 2 points here too: a. People may start getting crazy or something and asking about their money back AND the money of their burned stuff and b. DFI losses reputation and population and fans.....SO......NONE of the above will happen for sure!...... :DQuote:
Originally Posted by Dissolved
IF now it's a software (bios) problem, they are going to fix it with one way or another BUT better fix it quick coz they are the ones who loses of this whole thing...... ;)
I must ALSO say that MOST of the "unfixed" problems are hardware related - NOT only DFI mobos but all of them - and they are fixing them with .....another revision.........in a "silent way" you know.....;)
If that is the case it can´t be fixed with bios updates or anything right? And yes I do realize this is just speculation :).
Yes you're right........and.....Of course and it's just a speculation...... :DQuote:
Originally Posted by oqvist
Well i doubt DFI would recall hundards of mobos. so it looks like there gonna go the slient route, and get flooded with rma's.
The mobo was insanely priced when it came out, so they better do something, cuz i wont be happy if my board kills my good opteron.
I wouldnt mind going to another mobo, but i guess in terms of overclocking your always taking risks every second your above stock.
Quote:
Originally Posted by hipro5
But this issue isn´t an overclocking issue is it? It sounds like it can happen as easily in stock settings as for overclocked ones...
Be a little patient? Wait for a new revision? I've already bought this board and it's not is if DFI will just give me a new revision, one doesn't exist at the moment and may never exist. I just don't have the money to play this kind of wait-until-my-sh*t-catches-fire-and-dies game! DFI appears to have a lemon on their hands, at least that is my opinion, so as not to get flamed by members who 1. Have this board and 2. Been lucky enough so far to not have any issues.
Shawn
i want a new bios for Christhmasssssssssssssss-.....
FYI, the 12/7 BIOS is back on DFI's website.
http://us.dfi.com.tw/Support/Downloa...FLAG=B&SITE=NA
anyone know when the official bios is coming out.......i ain't beta testing for DFI that's for sure
I believe there are more elegant ways to supply multiple voltages. Its easily accomplished by means of a programmable voltage regulator. You have a transistor-mirror configuration, where the base can be programmed, with the output coming off the collector. Simple, easy and elegant. Resistors have to have a high tolerence over a wide range of temperatures and the inductive fedback due to switching may cause problems. These resistors are usually metal-film type (carbon film has poor temperature coefficient), so there is always an associated lead inductance.Quote:
Originally Posted by saaya
I must say I don't know how this is controlled on the board, hence the speculation on my part. :)
I think you are referring to a slow-start ckt, which is very easy to implement and can be quite robust. I don't see why one should see a sudden surge/spike if a slow start ckt is properly implemented.Quote:
and this is what might actually have happened on the boards.
when you boot the board the bios configures all the voltages on the board and sets them to what it thinks is default. so no hardware gets damaged, the resistors probably get configured before the circuit is actually powered and produces vcore, so that the first vcore the cpu gets is some good clean 1.5v or 1.4v, whatever its default vcore is. same for the other voltages.
-BIOS chip problem can be sorted out with a new BIOS. But there could be an unintentional leak or cross-talk (traces being too close, but that is highly unlikely), causing the BIOS to garble up.Quote:
i think this is the "default values" dfi mentioned get lost, or can get lost.
the problem could be several things:
-the vcore circuit gets powered up before the vcore is configured correctly, "just" some bad timing
-the bios doesnt know anymore how to configure the circuit because the data was erased or corrupted
-OR, its not the bios after all! its the chip that is supposed to set the right voltages that fails and all voltages get set to default or to random settings resulting in voltages way out of spec for the cpu and maybe vdimm as well (could be that the voltage control chip that dfi uses and that was burned on one of mikeguavas boards controlls vdimm and vcore)
this is all speculation, i dont know sht about designing circuits :D
i dont eve know how to find a vmod on my own :lol:
and i just got home from nightshift and worked from 9pm to 6am so forgive me if i wrote up some bs :D
im just trying to make sense of all the things i heard about this issue
-Bad timing you speak of could be a crap slow-start ckt (if its present).
-I didn't know that one single chip was used to control both VDRAM and VCORE. Thanks for the info. :)
[/QUOTE]Quote:
EDIT: pcdog, the second cpu you put in fried the power circuit, right?
and the same thing happened on mikeguavas board...
now was that second cpu you put in a 130nm cpu?
if yes than this is very interesting.
mikeguava also used a 130nm cpu in at least one of his boards, if it was the board that got the fried power circuit thats interesting.
130nm a64s have a default vcore of 1.5v, 90nm 1.4v
if the voltages get set to default for some reason the vcore could be higher for 130nm cpus than 90nm cpus, hence the circuit burns out with a 130nm cpu but doesnt get damaged with a 90nm cpu and only kills the cpu. if the vcore that gets apllied after the default settings have been lost, that kills the cpus, is actually higher for 130nm cpus than for 90nm cpus then the power circuit doesnt completely fail but still works to some degree.
hmmmm this would actually point to the vid special control wich lets you increase the vcore in percentages. its using the normal vcore as base. the base vcore is higher for 130nm cpus, hence the resulting voltages with special vid applied are way higher on a 130nm cpu compared to a 90nm cpu.
so if it turns out that 130nm cpus kill the boards if the issue happens and the default settings get lost somehow, while 90nm cpus dont and only get fried, then it would mean the special vid setting is the problem.
If programmable regulators are used, I think it usually waits for a signal regarding the default VCore before it can apply it. Now, if that signal were garbled, anything could happen. I would agree with you regarding the VID special control, but I also believe that in mikeguva's case it would have fried up, regardless of the CPU type. It takes a whole lot to toast a CPU! In fact 130nm CPU's are a bit harder to toast than a 90nm CPU. Something must have gone really really wrong there. :stick:
you mean the heatsinks on the mosfets are grounded, right?Quote:
Originally Posted by hipro5
and those sticks that hold the heatsink on top of the mosfets are too close to the mosfets so it can cause a short circuit?
Super Nade, i didnt get the part about creating different voltages in an elegant way ^^
can you explain it again?
so this is most likely how boards create vcore and vdimm?
and could this still result in the voltages beeing too high if the circuit is not configured and runs at default settings?
or would you have to miss-configure it to get a high voltage?
about one chip controlling vcore and vdimm, thats speculation!
i know that there are chips that can control multiple voltages, some videocards have one chip that controlls vddr and vgpu. i just speculated about it as it would make sense with MFMs memory dieing.
hmmm but that could also have resulted from a voltage spike comming from the cpu when it died...
and yeah, i think that mikes dead 130nm a64 was killed by a board that had already been damaged when it killed a 90nm a64 before.
or he had the special vid set to very high and maybe that resulted in an even higher vcore getting supplied to the cpu.
thx for your input nade and hipro!
good to see some experts opinion about this :)
hope you stick to this thread :)
Beta BIOS is up on both global and North America.
http://us.dfi.com.tw/Support/Downloa...FLAG=B&SITE=NA
Well, I through in my cheap chip and have been banging on this board for two days now. I've tried every combination of mem and vcore voltages and rebooted probably a few hundred times. I just can't seem to reproduce anything. Man, this issue is something else.
I did have one wierd occurence though. Maybe it is a clue. I put the system into S3 standby. Upon awakening, all my temps and voltages were screwed up in Windows. When I went back to the BIOS on reboot, all was well. Not sure what that was about. But, in Windows after the stand by my voltage was showing .1 too low, and my temps were WAY high. Strange. I'm not sure if the reporting was accurate or not.
I put my FX 57 back in at this point and I'm priming away at 3.45ghz. I wish DFI would get something official posted soon.
Quote:
Originally Posted by mdzcpa
i have the 11/25 and my system is flawless. i don't want to get another mobo. so im hopeing that if i dont fool around with my system too much, DFI will get this issue fixed with an offical bios.
Congrats - you just found ONE of the bugs!Quote:
Originally Posted by mdzcpa
Quote:
Originally Posted by Dissolved
I understand. I'm just letting folks know its out there as there was a report it was removed...which isn't true.
I had zero problems with the official BIOS as well. I was running the shipping BIOS. No issues on two boards.
I went over to a few forums to see more reports (HardOCP, Overclockers.com, AMDMB, techreport, anands, OCforums, etc) and the reports are very scarce. It seems most of the reports are centrlaized here. Even DFI Street has nary a mention of trouble except for links to XS.
I dunno. This is getting frustrating...DFI needs to say something.
LOL...I would have found it sooner, but I never ever use stand by :) I've just been trying everything I can to duplicate the issue and get some clues.Quote:
Originally Posted by mikeguava
If the CPU is not getting VCORE until POST but is getting voltage from the memory, wouldnt that be more or less suicide considering the other thread here which is discussing how low vcore vs. high vdimm kills chips?
Quite possible. Good theory coming together so far.