"Tiny matter that bears no impact on anyone", who are you kidding? Some people here got screwed out of a month playing their favourite game at 100 frames per second, instead of the 90 they're getting now. That's a damn crime.Quote:
Originally Posted by Particle
LOL...
Bunch of kids thinking they are smart at they own jub, and did it all wrong and that expirience cost amd lets say 200k dolars? or more?
Haha, you mean each and single 6970 card is nerfed from 1920sp to 1536sp by manually sawing a part of the die off? :DQuote:
Originally Posted by tbone8ty
Sounds more like Nvidia GTX480 story! :rofl:
I actually hadn't thought of it that way. I'm not immediately sure how I feel about that angle. Let me think about that.Quote:
Originally Posted by freeloader
amd taped out both 1920sp and 1536sp. they probably found 1920sp too hot like fermi so they went for 1536sp and saved the crown for dual core.Quote:
Originally Posted by zalbard
Or they only planned 1920sp part for 32nm, thus it's never been taped out. You never know, unless you work at AMD.Quote:
Originally Posted by sniper_sung
ROFL of the century. That's just amazin :D
roflQuote:
Originally Posted by kaktus1907
Woodscrew stories are nothing compared to that.
That's customer's cards we are talking about.
Don't know if it's the real cause of the delay but that's crazy to see that kind of mistake. AMD puts on his shoes before his pants on this one :(
looks like someone at AMD doesn't know how to use a ruler... :rofl:
they didn't even notice the problem in solidworks? oh wait, solidworks uses cuda, nvmd...
Yeah, I'm sure that because solidworks can utilize cuda AMD engineers are relegated to measuring and drawing everything by hand.Quote:
Originally Posted by 570091D
:rolleyes: :shakes:
If they were made of gold i'd have filed them all down myself.Quote:
Originally Posted by XSAlliN
(And kept the shavings as payment).
Even the thought of it is flipping hilarious! :lol:Quote:
Originally Posted by Solus Corvus
What? nVidia lying to customers and partners about having a real card when all they got is a fake is nothing compared to a changed cooler being at most 2mm to long or wide?Quote:
Originally Posted by Olivon
Lies and faked card is a bit worse than modifying a connector by hand. It doesn't affect function or reliability.
im suprised noone has mentioned yet but here is the AMD's response..
Quote:
The modification to the power connector seen on some Radeon HD 6900 series cards was a conscious development decision. The reference boards use both fan and heatsink designs from multiple vendors and it was discovered some material produced on initial designs required adjustment. As a result we modified the PCIe power connector for a very small number of boards using affected components, which were mostly used for samples, as the most time and cost efficient solution to bringing the product to market in volume. This minor modification to the connector in no way affects the operation of the card and our partners have designed solutions which do not require this modification.
Means most retail cards won't even have this. :)Quote:
As a result we modified the PCIe power connector for a very small number of boards using affected components, which were mostly used for samples
Here in France, first HD 6970's owners got the PCIe connector modded.Quote:
Originally Posted by zalbard
In Germany too, here are some pictures with a friend's retail card...
It was interfering with fan's blades.
http://forum.lab501.ro/attachment.ph...9&d=1292595554
http://forum.lab501.ro/attachment.ph...0&d=1292595559
http://forum.lab501.ro/attachment.ph...1&d=1292595566
arent cards going into some sort of industrial hoven to solder the joints or whatever and too much heat shaved the corners or whatever???
am i the only one seeing this possibility????
and did it hinder the performance of the cards ... so why complain people ...
Because it's an huge fail, simply ...Quote:
Originally Posted by Sn0wm@n
Quote:
Originally Posted by Olivon
did you read anything else .. or you just bothered to scream fail ....
Is there the same problem with the 6950 ?
Which brand have this crude modification : only sapphire or not ?
A huge fail is when the fault is found by the customer and there needs to be a recall or remedial action. This one looks like in was found in house.Quote:
Originally Posted by Olivon
That's not true. A guy here in Portugal bought two 6950 and both had the 'modded' connector.Quote:
Originally Posted by zalbard
You didn't see xdan's post ?Quote:
Originally Posted by Sn0wm@n
Without the modification, the fan wouldn't turn because it would be in contact with the plug.
You see ?
Did you, or any user that purchased one of these cards, have a problem with the fan or cooling? Seems like they took the bullet on that one for the customers.Quote:
Originally Posted by Olivon
--Matt
Doesn't the 6950 have two 6-pin connectors (instead of a 6 and 8 pin) - so is naturally shorter and shouldn't interfere with the fan :shrug:Quote:
Originally Posted by jeanjean15
it would be fail if they didnt fix the connector so the fan didnt spin. considering the card works 100% like its supposed to i dont see why everyone is crying fail
Yeah, true. BUT they found this at the factory, and fixed it there, so it works fine. It isn't like they waited until it was in the retail channel and went "oops" then. They took care of the problem before it hit retail.Quote:
Originally Posted by Olivon
While it is a measurement fail on their part, and a bit amusing, I do think some people are getting overly worked up over nothing.
I just don't understand people saying that this is not a big deal. It reflects very poorly on AMD. In fact I am astonished that such a large company does something like that. But since many people don't seem to understand why, let me explain.
AMD don't manufacture the cards themselves they subcontract production. Before any subcontractor start full production of a product they require the signed approval of pre-production samples. Pre-production samples are always made and are a mean to catch any late defect in a product.
In other word AMD would have approved the samples before mass production starts. The only way that this would not have happened is if AMD had agreed to start manufacturing without approving samples, which frankly is unthinkable.
Now, as always, crap happens. And it is possible that for whatever reason the subcontractor screwed up after samples were approved. It is rare but it can happen. 99.99% of the times when this happens, the whole batch is rejected for not conforming to the agreed specifications.
The subcontractor is then responsible at his cost to correct the problem and very often they have to write off the whole batch. And a customer of the size of AMD would most certainly had strong-armed the subcontrator to prioritize its production and/or face monetary penalties.
But what happened here is quite extraordinary. The whole batch was manually "modded" so that the cards could be assembled and sold! This can only mean that either:
AMD was extremely late and had no time to wait for replacement cards (a little doubtful when you see that this was certainly a very good hard launch).The cards could be made on time but it was too late for sea shipment, and AMD would not pay for air shipping (which is usually what most manufacturers would have chosen in this case)AMD decided that most people wouldn't notice anyway and the risk of not having a card to compete head on with the GTX570 for the holiday season was just too high
What this shows as well is a lack of respect for the AIBs. And I would be very interested in knowing if the largest such as Asus or Gigabyte have accepted to take delivery of this batch. I wouldn't be surprised if they rejected the cards and decided to wait.
As far as the end user is concerned, although the risk of this mod making the card defective is low, the risk is definitely there. In addition, this could potentially create an absolute mess if one or two years from now the card become defective. And the user will have to trust that the retailer and the AIB will have notified their tech support that this card was moded by AMD, and not modified by the user.
I have nothing against AMD, but this is very poor. If anyone want to purchase that otherwise very nice card, I would recommend to wait for the revised card that will certainly arrive shortly.
Sadly not .Quote:
Originally Posted by sam3
It is exactly the same problem . looks at this :
http://hfr-rehost.net/fullsize/http:...3/frontiwb.jpg
+1 for being blown out of proportion
the only thing to be angry about is delays.
Isn't Sapphire ATI's biggest partner? Same with EVGA, nvidia shafted the rest of them all the time.Quote:
Originally Posted by Hannibal Lecter
In fact, they shafted BFG into bankruptcy, XFX went ATI and got cut out, etc, etc.
A SNAFU that probably required cooperation from everyone to fix is not the same as strongarming.
They may be their biggest partners but Gigabyte and Asus are larger companies and I would be surprised if they accepted to take delivery of these cards. It is pretty easy to check if retailers have Gigabyte or Asus cards in stock.
I don't know what nvidia has done or not done and this is not the subject, what AMD did, reflects very badly on them. I don't think they expected that people would find out about it.
And for the record I would have said exactly the same thing for any other manufacturer.
Its not about the product working fine, or getting worked up over nothing, when a company as big as AMD lets this slip by, one has to think how it even was possible to let this happen, and the fact they shaved it down, mickey mouse style, and still sold it to the customers is bad business in my book, if you have no problem with it, then thats you. AMD should have fixed this, taken the loss and re print new cards. You just dont do something like this when you are a billion dollar company such as AMD. :down:
AMD made an official announcement saying only a very small portion of the cards were affected.
So far, modded cards have been sold in France, Germany and Portugal, cause users have reported it.Quote:
Originally Posted by hurrdurr
It does reflect somewhat bad on AMD, But it would be even worse for AMD to send back the boards and cause a 1+ month delay and miss the xmax buys. MOST people that buy these cards are never going to take off the heat sink cover. As long as it works no worries. But I am sure after this AMD will be triple checking their future designs.
If it was a small portion of cards, they should have had no problem taking the loss. Instead they decided to grab some nail files and got to work :ROTF: Shame on AMDQuote:
Originally Posted by hurrdurr
We are a lot in France having this defect so i think that AMD do not tell the truth .:shakes:Quote:
Originally Posted by hurrdurr
+2, I don't think its a big deal either, amusing but not a big deal.Quote:
Originally Posted by Manicdan
As for the issue itself I would hazard a guess that AMD reworked every card in the batch even though only a small amount were probably affected. With various tolerance build ups, product aging and slight movements over time they probably played it safe.
if it was a small batch, im actually surprised they even delayed it at all, they should have just launched it a few weeks ago, with 10% less stock on the market, and let the second wave of shipments include the problem ones after being fixed.Quote:
Originally Posted by Easybeat
in the end either option would have had the exact same number released as of today, just in different quantities.
Maybe AMD was surprised by the GTX570 and they had to do some last minute changes to the card. But even that makes little sense, because the heatsink would need to be installed so they would have caught the defect.
So it's probably a subcontractor screw up, and they decided to still ship, and they also probably got a sweet deal on these cards.
But seriously, this batch of cards could become a big guaranty headache. I personally would wait before buying this card. Why take the risk?
This sort of thing happens and I don't think it's big deal,it doesn't affect functions of the card.My theory of how it happened:when you get a lot of parts from a lot of different manufacturers,miscalculations happen,for all we know the team that designed the card didn't account for fan vibration during operation,simply because of miscommunication with the manufacturer of said fan. As simple as fan is,that detail is easy to overlook even for an engineer ,when it's not the your area of expertise.
Also AMD doesn't exactly make these cards,a third party manufactures them,I am not certain about the details but they could be designing the cards(or parts of) as well.
And these things happen a lot and I mean that,any manufacturer is prone.Car makers,electronics makers,in construction when faulty designs get overlooked/approved and noticed/fixed/rigged at the last moment or not at all... but in this case it is moot,it doesn't affect function.
Wood screws is a whole different situation and shouldn't be compared,but when nvidia made thousands of faulty chipsets,that keep dying to this day,that's a situation where design flaw actually matters because it impacts performance,that's why they are getting sued.
Whether or not this was the cause of the delay remains a mystery to me,it could be either way.
You are absolutely right this can happen, and a badly designed product could deteriorate faster than expected. But this is not what happened here. They had to know very early, since the heatsink cannot be assembled.
AMD say that only a small number of cards are affected, but if it is true, this is even worse: why not scrap these cards? It doesn't make sense at all.
We all know that all manufacturers at one stage or an other will face this. What is wrong and surprising is how AMD chose to handle it: surprisingly poorly.
No,there is in fact more then one way to make something,this is an acceptable way,there is no reason to scrap the cards.
how exactly can this cause a product to be more degraded? its metal prongs are already exposed anyway
There is a reason for the plastic around a connector. But you're right, the risk is very low, I would be more worried about the potential RMA problem in case of an unrelated later break down. The card should be fine. Just like a TV sold with scratches on the back will most certainly work fine, but not what you would expect, even less if it wasn't an accident but with the knowledge of the manufacturer.
Quote:
Originally Posted by RSC
Remember a "small portion" of cards could still be 5k or 10k cards scattered across the globe. Of course you are going to hear people reporting "hey, this is filed down" way more often then "hey, my card is 100% normal".Quote:
Originally Posted by HelixPC
Because of ? Oh yes, I forgot, the pesky "PCIe power connector implodes due to a lack of a supporting corner" syndrome that we've been suffering from since never. Electrically and functionally these cards 100% fine. Aesthetically there is a slight nick in each one. I'd have rather they used these cards as RMA warranty cards than original retail cards but the filed plastic won't impact the lifetime of the card for the normal user.Quote:
Originally Posted by Hannibal Lecter
Just hypethetical, If lets say for what ever reason that connecter was designed to have an area shaved off(done in the mold) Would it still be an issue to everyone....Probably not.
Because the store/retailer/brand can say you did the "mod" and because of that your warranty is not valid.Quote:
Originally Posted by [XC] gomeler
It's a hardcore scenario, but it can happen.
Quote:
Originally Posted by [XC] gomeler
You misunderstood me. I am not saying that the "mod" will become a source of failure. I am saying that if the card fails for whatever reason, it could become a problem when tech support at your retailer or the AIB inspect your card for signs of tempering before considering any repair. You would have to trust these people to remember about this batch or to even have been advised about it. If it happened in the next 6 months you may be fine but what about in 2 years time?
Why would you take any additional risk for the mistake of somebody else? Why not wait a few weeks before buying the card? That's what I am saying, hopefully it is clearer now :)
It has nothing to do with waiting. It has everything to do with the cost of scrapping the cards!
The delay has less to do with this anyways, than it does with the 32nm debacle.
Is this somebody know if the gigabyte cards have got the same crude modification ?
This oughta knock the ATI fanboy's down a peg. Unfortunately it also raises the Nvidia fanboy's 2 pegs. Wonder if there's a file for that....
And you can't even see it unless you specifically look inside the card using a flashlight...Quote:
Originally Posted by [XC] gomeler
i dont care, i would file it myself when i do a cooler mod :)
i find its a good move, they have more cards in stock earlier is what counts in for sales. If they did a good deal with whoever is responsible its even more clever..
I agree with Oese. I don't care about the connector corner being hand sanded off either. You can only see it if the card is out of your case and you look really hard at it. It doesn't effect the performance of the card. If any of you had ownership in AMD you would have sanded it to. I am sure it is more than lets just re-solder this connector over here. They would probably have to do some redesigning to the PCB board, moving components around. That would have required too much time initially and also would waste perfect boards which means a lot of money wasted. NVIDIA already had their gtx series out and AMD needed to get theirs out on the market. I could care less about the corner. Mine work perfect.Quote:
Originally Posted by Oese
I don't see why anyone is complaining about this. What if they didn't sand off the corners of the connector. Then you guys would really have something to complain about. What if they initially scrapped all those boards and made a revision board. That would have taken months of time to do and yet everyone would be still complaining. They definitely made the ideal choice, both for AMD's business and for the consumers who were eagerly awaiting the new 6900 series.
they couldnt just make a different cooler?
look at date of the other posts
Yeah they may have been able to do that, but the best course of action was to simply just file the connector end. To make a new cooler, first it would have needed to be designed, then from paper to the plant to be manufactured. Those things take time to complete. They might not have had all the materials in house and would have to order them, and then, they would need time to test the cooler (and the cooler performed as they wanted to so why change something that already worked), and etc... Not only would that have been time consuming, but also costly to do. I am an engineer but you don't require an engineering degree to figure this out.Quote:
Originally Posted by Aleki
Could anyone confirm whether the MSI 6950 Twin FrozR II being affected or not?
http://www.overclockers.co.uk/pimg/G..._34803_400.jpg
They could have saved the time on mine. Mine is water cooled as of about a week ago, Yay! I have a sapphire 6970 with the filing and I'm not concerned. I'd be surprised if any of the non reference cards are "affected" sniper sung but again why worry about it it's just the plastic that is filed a bit and not even to the point where the pins are exposed: it's through one of the 2 walls.
this doesn't affect performance in any wayQuote:
Originally Posted by sniper_sung