4870 dying thread

[AliG]

I don't own a 4870 yet, but I plan to, and someone else pointed out that we should do a systematic way of figuring out what the main cause of the deaths are. I myself believe its the memory, but just to be sure, please chose the appropriate answer in the poll to pinpoint the problem.

thank you for your cooperation


For all people that did flash your bios, there is now a fix for your dead cards. Yes you did essentially kill it by your flash, but you can make it work again with this fix
http://www.techpowerup.com/articles/...ng/vidcard/152

For those who didn't flash (and once again, you know who you are if you flashed your bios) and card still died, go ahead try flashing the bios anyways, a lot of the time a simple bios flash can fix most errors, if it still doesn't post, you'll have to rma your card

[AliG]

ohh and please be honest if you did flash your bios, it's not helping anyone if you did and you say you didn't

[bill_d]

need a card is working fine and if you have oc in ccc

[Marvin_The_Martian]

Like I said in the other thread, public poll? With your second post? You should know better lol! Maybe, an rma is easier if it's not your fault, so who would publicly state they done something to kill it themself...

Change the poll imo.

edit:

need a card is working fine and if you have oc in ccc

I'm sorry, first off all

Now what was it you were saying?

[jimmyz]

great Idea. I was going to pm everybody and ask them in private whether or not they flashed or O.C.ed. don't know why I didn't think of the poll idea.


PLEASE BE HONEST, LYING WILL ONLY DELAY A POSSIBLE FIX

If possible give any other info regarding the death.

so far the typical scenario is this:

1. Card works fine stock , people play for a while then decide to OC.
2. OC in CCC seems fine and cards scale mostly to the limit of CCC.
3. People decide to flash, either to increase fan speed or clocks, seems either is not working.
4. Card runs fine for as long as the unit is powered up, most people even reboot without losing them.
5. Eventually the card is powered down completely, as in no power to cards for a few seconds to hours.
6. After powering completely down card refuses to boot. and actually hangs system.
7. Three of the four lights stay on and fan spins 100% (or at least sounds like it).
8. People who are able to install another adapter report seeing the card both in windows and dos, but flashing does not correct.


If anyone else has experienced a different scenario please explain as much as you can about the circumstances.

[bill_d]

needs more options
1 oc in ccc card still working
2 stock card still working
elc.
if you want do get a % cards good to bad and what was done to them or not

[SNiiPE_DoGG]

i was honest that i have one card dead without flashing or OC ing but i have another that died a few boots after flashing... sorry i cant represent it in the poll


ALSO JIMMY: for the card i flashed that is the exact scenario that happened to me, i flipped the switch on my psu to plug in my cd drive's molex connector and after that it was done

[AliG]

this isn't about the card working, this is about the card not working

ohh and reason for a public poll is so we can talk to the people that didn't oc or flash their bios and hear what they have to say, this is about pinpointing a problem, get over it as this will help more than take from people's privacy

[Marvin_The_Martian]

this isn't about the card working, this is about the card not working

ohh and reason for a public poll is so we can talk to the people that didn't oc or flash their bios and hear what they have to say, this is about pinpointing a problem, get over it as this will help more than take from people's privacy

There is talk about getting a percentual view of this occurance, for which we need to know how many dead and how many working been mentioned in the other threads and I think it's a very good idea.

Edit: and 'get over it'? Wrong foot this morning?

[AliG]

sorry, there's just been so much flaming in the news section lately its hard to tell when people are being blatently stupid or are actually trying to help. Something really should be done about that, but I guess there's only so many people you can ban before you start to take away too much from the forum

[Shadowjump]

What BIOS are you talking about guys?

[Marvin_The_Martian]

sorry, there's just been so much flaming in the news section lately its hard to tell when people are being blatently stupid or are actually trying to help. Something really should be done about that, but I guess there's only so many people you can ban before you start to take away too much from the forum

Yeah nm I already explained why I would make it private in the other thread, it's just to make sure people are honest and not scared their rma will become harder since their user details and actions are in plain view. In the end we all want the same, answers to questions and solutions for problems.

[SNiiPE_DoGG]

Ok per request of Ali here is the complete and honest accounts of my cards deaths (individually)

Card1: i had installed the card and booted into windows and i proceeded to begin downloading 3dmark Vantage (new install of windows ^^) while i waited i used ATItool to test the load temps of the card and run an extended artifact check. everything was fine. I started up the game 'Project Torque' in order to give the card a little go, nothing super intense. after closing the game i was messing with ATItool again, but found nothing useful; i started another artifact test. This time in my card created many artifacts every ~3 seconds and had major issues. a few seconds later i started getting artifacting all over my desktop in the form of horizontal multicolored lines. this worried me greatly so i shut down and upon restart the artifacting was occuring on over 50% of the creen area at all stages of the boot and on the desktop. there was no OC, no bios flash, nothing, the card is most definitely dead.

Card2: after my first card had died i put my second one in (didnt have the powwer in my PSU for both and was too lazy to install my new one) it worked great initially, i used ATItool to test the cards temps and decided that my load temps were too high for stock, i used RBE to edit fan profiles for the card ONLY and restarted and reflashed many times, trying to get it just right. Every time i flashed i used a fresh modded bios, never building on the previous one. at this point i had great temps (load 55c ) so i began to OC, i sucessfully reached 790/1060 and decided to leave it there to be safe. i went to go install COD4 but to my dismay my DVD drive had its power connector knocked out . i then proceeded to shut down the computer and power off the PSU. upon restart my system would not post with a black screen, my DFI poster LCD reads code 26 when this occured. If i leave the system running at code 26 the fan speed on the card appears to change randomly, sometimes very high for a few seconds then drops to <10% the doddle around between 10-100% never repeating the same speed pattern. i get no LEDs lit up (i think, i never noticed them when the card was in the machine)


so theres the story of two cards dying in completely different ways, hope this helps

[Marvin_The_Martian]

What clocks did you reach, especially for the ram on card one? Sounds on this one the ram is the culprit.

[SNiiPE_DoGG]

What clocks did you reach, especially for the ram on card one? Sounds on this one the ram is the culprit.

card one was at stock speeds for 100% of the time i used it, never oc'd

[Marvin_The_Martian]

If it died without ocing at this moment I would disregard that death to the sure to happen few mellons untill more people have come forward with the same issues.

[AliG]

hmm, leads me to continue with my theory that its a problem with the gddr5 ram, as no one with a 4850 has had this problem

[Marvin_The_Martian]

How many have been sold? How many have issues? Need to know this before we can make that assumption, even if it's an easy scapegoat. Probably it will be the problem, but we need to know when does that happen. when I mentioned this on the folding forum, they told me they tested this card before release and didn't have any problems with them period. What changed? I know for a fact they would not overclock, not even an mhz and certainly not while testing new clients.

So again, there is a need for more information.

Edit: MrHouston from the Folding forums has informed me their testing deparment did also test ocing the 4870 and they had no cases of dead cards. I'm hoping it isn't people trying to get out from under the fact they flashed a bad bios, which according to Dave Bauman in the other thread seems to certainly the case..

[xMrBunglex]

both of my 4870's work. update your poll and i'll vote.

[AliG]

How? I checked the thread tools and also the advanced edit

I think I'm going to buy a 4850 now, I don't want to take any chances not to mention I can get it for only $150 after 2 rebates...so yeah, that seems to be the best option for me at least. Now to wait for rivatuner to allow me to adjust voltage!

[Marvin_The_Martian]

@Ali: you can't change it after it's been made.

And I wouldn't start a new one just yet reading the last posts in the other thread.

You know what, I think we been lied to. Maybe I can't proof it, but if the number stays this low after a day of getting it under the attention I don't think it's really a manufacturers fault. Now Wizzard admitted the bioses on Techpowerup where all bad, it's just became more likely that the people saying they did not flash are not being honest. Offcourse it's just a gut feeling atm, but the same kind of gut feeling I had when I did believe them and if I could be vocal about that while some other respected members here already concluded it was from a bad flash I shouldn't keep quiet when I realize I been had.

When this settles, and it turns out people leid here maybe just to try and secure an rma I will ask them to be banned as I think it's the lowest conduct one can show and especially xs should make it clear it does not condone such actions from any member of it's community. I gone out off my way trying to defend the issue that if it's being reported on XS it's not without a doubt a case off user abuse.

[mr2sw21]

So Marvin you mean the product is fine just people flashing the bios is the problem. So if everything stock shouldn't kill the card?

[Marvin_The_Martian]

I'm saying the folding community has tested the 4870 extensively before it ever came out, both oc'd and stock and never had a single problem. Now it could just be a bad batch, but seeing how incidental this seems to be since I can count just 9 dead cards ( iirc ) even that is starting to sound less likely.

Don't try to put words in my mouth thank you very much. Like everyone else, I can also talk for certain about things I know, this isn't one of them so I say what it looks like to me nothing more nothing less. If you are one off them who had a card die without flashing, where did you post all your details so people can figure out what's happening? If there were some more people doing that then the select few now, it would make it look allot more plausible that a hw failure is behind this then the situation at this moment warrants me to believe.

I fought against the notion that since this is XS, the people saying they died suddenly were lying without giving it a second thought, and it will bite me in the behind if they were lying. Don't fall over me being upset if they lied, I will try to get them banned if that turns out to be true, if only to show the others XS does not condone this behaviour just as any other forum. Lying about how your hw died, presumably to assure an rma, is the lowest a oc'er can do imho, well right after cheating scores.

[AliG]

yeah I agree this looks like bad flashing to be honest, that or they oc'd their card too much (as has been mentioned before, GDDR5 is quad channel, as small 100mhz oc is actually 400mhz ddr and that pushes the ram beyond its limit, most likely the issue is not enough voltage fed to the ram if that's the case or the ram overheated and died) by flashing a hacked bios.

[ghost101]

http://www.techpowerup.com/articles/...ng/vidcard/152