Topic: [WTS] Crippled 5970s - page 3. (Read 4031 times)

Well, I'm pretty sure I can get a much better deal on the 5870s.

As for the shipping, remember this is being sent from Europe, not the US. I might be able to get a cheaper price, as the ones I'm quoting are from the local mail agency, but for these weights it gets really expensive to send to another continent. That's one of the reasons I was trying to push this to someone in Europe, as with the same price I can get a fairly larger profit, with less going onto shipping.

ill pay the extra 50 for shipping are you sure its that much for shipping my address is citcy and zip are southampton nj 08088 if that gives you a better refernec to go buy

5970s: I'll pay ~$105 for shipping, which leaves $120. That's $60 per card and you get the dead one as bonus. Just stating this here in case someone else wants to make an offer and wants to know the item price minus shipping.

for all cards: so I'd be paying an extra $50 for the 5870s to be shipped along with the 5970s, which means that 475 - 225 - 50 = 200, or $67 per card. That is not all that interesting to me, to be honest. For reference, I take 1300MHs+ from these cards.

well ill do like 225 for the 3 5970 but i would like the coolers sent as well and i think i would do all the cards if you decide to sell the 5870 for 475 or my orginal offer for the 5 5970 for 400

You don't have to, but I would surely appreciate it

But the original sell offer for the 4 (plus the dead one) is no longer in effect, so I guess the short answer is; yes.

Probably tomorrow I'll make my mind regarding the 5870's, just in case you are interested on these as well.

does this mean i need to make a new offer? since ur only ofeering 3 instead of the 5

Ok, I'm really sorry but I'll have to change my offer. I have:

- 2x 5970 with a single core working (each pumping ~370MHs @ 820/300, fan 80% keeps temps at ~55ºC)
- 1x 5970 dead. It passes POST and is detected by the driver, but then the driver crashes at the first --odgc. I have it as just the board, the cooler is now detached (I wanted to do another visual inspection). Can be shipped fairly cheap like this, as most of the weight comes from the cooler, but I can also send the cooler (though detached, not wasting any more thermal paste on this one). As an added note, I used the screws from this one on another board, as I seem to have lost my big'ol bag'o'screws. I will find it, but I may not have time to search for it for some time.

The other 2x 5970 that were supposed to be cripple have found a new breath of life. I opened them to re-apply thermal paste in preparation for the buyer and when I tested them they were working again. I thought it would be a passing thing but they both endured a good 10 hours mining @ 820 without any incidents. They might still be crippled, I'm certainly not confident on their health, but they are working well atm. If anyone wants to make an offer on these I'll consider it, but 5970s are the most power efficient GPUs I have tested, so these are the ones I want to keep.

If I do end up keeping the 5970s, I might have 3x 5870 working perfectly (*) for sale instead... I need to juggle the power requirements as explained before, so the less efficient must go.

(*) - 1xSaphire, 1xATI, 1xXFX - the ATI has a wonky fan, it makes an awful sound at low RPM and seems to stop and start regularly, but when at 80% is works great, which is where I keep it anyway. I have one potentially dead XFX that could donate the cooler after I do a good check on it.

PS: Of course there's a chance that either or both of the recovered 5970s go south again, in which case I'll just go back to the original offer. Since I will not be shipping anything before Monday anyway, I thought I'd just come forward with this new information and I'll update the thread as needed if things change.

There is no obvious damage to the cards, that much I know. I assume that, unless there's a very obvious component to replace I'll be better off just selling the cards to someone that will make good use of the working core, but since you guys obviously know better if there is something else I should look for, some component I can pop off and test somehow, I'd be happy to try.

You could take it apart and check for visual damage. If a VRM has blown, I suspect youd see that. Whether that is reparable or not, I would imagine there is a fair chance the PCB is damaged too but not sure. I also suspect you'll have a hard time finding replacement components, but it doesnt hurt looking.

hehehe

Well if your worried about wasting your time this is definitely not the project for you, I would say that there is a 2% chance it can be repaired


but you never know until you try

I'm (obviously) no electronics engineer, but from the way the cards behaved, especially when I could still use them with 2D but not 3D I would say the VRMs are almost certainly to blame. Of course, I don't really have a good way of asserting that.

Is that something I can fix "at home"? I have access to top of the line soldering stations, all the test equipment I can handle, and all the other gizmos at a friend's factory, but I don't want to waste their time unless there's a chance it can actually work, and I surely can't use any of it without proper guidance.

(poor/lazy mans reflow)

preheat your over to 385

remove anything plastic from the card

bake it in the oven for 8 mins

wait for it to cool and see if it works!

(you would be surprised)

reballing would mean to take it apart, take the chips out, clean them and resolder them back in by hand



and p4 man is probably right but at this point it really can't hurt to try and it is the small chance that it works it would be well worth it

reflowing is baking them in an oven. It helps if there are hair fractures in the solder or underfil.  But thats usually only the case on certain nVidia cards and a result of thermal cycling, not something Id expect from an ATI card that has been mining.
My guess is that you burnt out VRMs

Well, I did learn it and I've been told no lesson is too expensive, provided you learn it


None of the two verbs presented make sense to me... sorry. Reflowing make me think about wave flow soldering, which is NOT something I'll try, and reballing makes me think of basketball which, hmmm, there's an idea!

I'll do
400 for everything even you paper weight and also the stock fans effect if you hqve them still

damn man, thats a tough lesson to learn

Have you tried reflowing or reballling them to see if it fixes anything?

I mean its a long shot but its worth a try

Indeed that's true and I have since learned my lesson, proof being I have a handful of 5970s and 5870s that work flawlessly and have done so for almost a year now. But these have been mining on the good core without issues.

They will not work for gaming because the primary core is the damaged one. Most will not even give any image, and while one of them (at least) does give image, using any 3D will immediately lock or reboot.

This is a perfect example of what happens when you run your cards too hot for too long and with the fans up too high

I suspect we will start seeing many more threads like this in the months to come, Ive been mining almost as long as he has, the only card out of my 17 that I have crippled is a reference 6870 that does not like to mine clocked over 950mhz but other than that I have had good luck

I have never run my fans over 75% and now I have most of them around 50%


Question, does the card still work for gaming? You would probably have better luck selling it as a gaming card, coincidentally gaming takes much less resources than mining does

That is one great question. They didn't all die simultaneously, mind you, but I've been mining for over a year now and I have had my share of overheating cards, so I'm guessing that's the thing.

Nowadays I do either WC or open frame with pci-e extenders, but there was a time when I ran 2 or 3 5970 on a smallish case with improper air flow, meaning they would run at 80~90ºC for weeks in a row. So I'm guessing that did it, but they didn't just fail, the cores would start locking randomly (always the same core) and I would eventually take out the card, take the cooler off, apply better thermal paste and replace pads as needed, it would run both cores for a while again and then the failing core would start failing again until eventually the failing core would die and neve operate again.

There's at least one of the cards where the failing core is still operational, in the sense you can mine off it for a while before it locks. I have also completely killed some 5970s along the way, and I have one such paperweight if anyone is after the cooler or believes they can fix it. The card does POST and gets detected by the X driver, but one call to aticonfig --odgc takes a minute or so to respond, after which the driver locks.

Just how exactly did you manage to (half) kill 4 5970s?