Slashdot Mirror
AMD Graphics Chips Could Last 10X To 100X Longer
An anonymous reader writes "According to a research report out of UCLA, released this morning, NVidia's high-lead bump packaging could last anywhere from 1/10th to 1/100th as long as AMD's advanced eutectic bump approach. (TG Daily has picked up the claim.) NVidia is currently in the midst of a $200M recall of bad GPUs, and the report suggests that the issue could be much deeper than NVidia's PR department would have us believe." The report lends credence to the strident claims of the Inquirer's Charlie Demerjian, which we discussed a month back.
I'm all for longer life chips, but what are Grahiphics ?
I was going to Google for that word but then I realized that kdawson was involved.
The report lends credence to the strident claims of the Enquirer's Charlie Demerjian
As in National Enquirer?
As in Real news?
--
Oh Well, Neutral Karma and all . . .
In my other life, I eat cats.
"Silicon doesn't wear out; microchips were effectively immortal. The Wig took notice of the fact. Like every other child of his age, however, he knew that silicon became obsolete, which was worse than wearing out"
"I bless every day that I continue to live, for every day is pure profit."
What does 1/10th and 1/100th actually mean in standard solar days?
Can someone please provide a plot of the various solders and their performance vs. temperature and time?
I would like to see the plots for ====>
90Pb10Sn
60Pb40Sn
97Sn2.5Ag0.5Cu
99.3Sn0.7Cu
96Sn4Ag
99.25Sn0.75Cu
What is the risk associated with Tin? Especially Tin whiskers.
What kind of solders does the slashdot community use?
Nvidia has already switched away from high-lead solder.
The Inquirer reported on this whole fiasco.
People shit on The Inquirer a lot, but there are 3 awesome things about that site:
Their writers do not sign NDAs.
They have writers all over the world - not someone they send out, but people who live there.
Their writers intimately know people in the industry - from the people up top to the people at the factory floor.
This study does NOT specifically address or study AMD or NVidia's Chips.
It does not specifically address or test the exact chemical makeup of chips belonging to AMD or NVidia.
The conclusions being drawn as to the relative life spans of those manufacturer's chips appear to strictly belong to the bloggers who want a big headline, and not to the authors of the study. The study authors specifically note that in order to determine the life span of real chips, the real chips in question should be studied. Quote:
"For life-time prediction, the real microstructure of these two kinds of flip chip solder joint should be studied and actual failure rate should be measured. "
The study states that they are ignoring various factors that would come into play in the real world in order to simplify the study, and that they are making a number of assumptions about various testing conditions and about the makeup of the materials themselves.
From reading the study linked, it's not even clear to me that they actually tested anything, and it appears from their wording to be only a theoretical exercise.
In no way should the results of this study be used to state that brand X's chips will have a longer lifespan than brand Z's chips.
And as you tread the halls of sanity, You feel so glad to be, Unable to go beyond. I have a message, From another time..
The chips on my cards have always outlasted the fans on my cards. I have owned both nVidia and ATI cards.
Just because the chip - or at least, one aspect of the chip *could* last longer doesn't mean the card will.
I have an older 9550 ATI in one box and my more powerful machine has an 8800 nVidia in it.
As far as large, clunky drivers go, ATI is king of the hill... their setup that requires .Net to install, the bloated and resource hungry Catalyst Control Center... it's ugly.
nVidia on the otherhand seems to be far more lightweight and fits in better with Windows.
But performance wise I haven't really had anything to complain about though, and I can't think of instances of actually cursing the drivers for not working...
I've never been loyal to either really, when it comes time to do an upgrade I research on the web what card people are saying gives best bang for buck in my pricerange, I don't give a hoot who makes it.
From TFA:
"£GDl/h' = 13.5/10. Clearly the strain in the eutectic SnPb layer (in the composite solder joint) is about one order of magnitude higher than that in the homogeneous eutectic SnPb solder joint."
What fucking dipshits! I can't believe those morons at NVidia didn't realize this. Any judge is going to take one look at this in a class-action lawsuit, and NVidia is HISTORY, man!
Fascism trolls keeping me up every night. When I starts a preachin', he HITS ME WITH HIS REICH!
I'm skeptical about the report, especially given the lack of any field studies with it. The useful life of a piece of computer equipment is usually 3-5 years; high-end graphics cards are probably shorter, because the main customers are gamers who need cutting-edge performance to kill orcs with.
So does "10-100 times longer" mean that significant fractions of nVidias are failing in 10 days - 3 months due to bad solder joints? Or does it mean that the solder joints in an ATI will last 30-300 years, long after anybody except a few retro gamers are interested in a graphics system that's mounted on a card in a separate box and doesn't interface directly to their optic nerves?
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
Most companies offer at least a year long warranty; if they have significant failures in that year, like 10-100x higher than normal, that may put too much pressure on their warranty policy.
And let's not forget nVidia's partners in selling cards (you know, all the non-nVidia nVidia cards). Those people may see high failure rates of nVidia parts, and all of a sudden using another chipset just got a heckuva lot more attractive.
So, the moral of the story is, there is no set 'time' that a card will die. It's not like after 10 months all of them will just conk out. But if there are higher failure rates than normal in their warranty period, not to mention harm done to their reputation, it could end up costing them greatly.
Just because you don't like the truth, does not make it false.
Now I'll have to seriously consider switching to unleaded.
Invenio via vel creo
"Since the plastic energy produced in the eutectic SnPb layer in the composite joint is about 100 times larger than that in the homogeneous eutectic SnPb joint, we expect the cycle times needs to fail the latter will be 100 times longer. But the above model is ideal, nevertheless it is reasonable to say that it is at the least 10 times longer. " A) They didn't test ANYTHING. B) They admit their ideal model is probably not realistic, and actual difference could be much less C) They admit this does not mean actual life-time of products and explicitly suggest testing in that regard. D) /. needs to stop posting sensationalist crap like this. The article is interesting, the title is bullshit. If I want sensationalism I will turn on my TV.
"Dictator Flakes. They WILL be delicious."
Comment removed based on user account deletion
How to fix bloated ATI drivers :
1] Download full CCC installer.
2] Don't install the CCC, just the drivers.
3] Download a copy of ATI Tray Tools.
Big! Strong! Wow! Tada-O!
After the dive AMD took today in the markets, they better be long lasting, because it's going to be hard to find replacements.
To do list for Windows
With the lead content, they're not RoHS-compliant.
Typical failure models use an exponential distribution, rather than a Gaussian distribution to model time-to-failure.
I can confirm the putting pressure on the warranty part. Dell just ran out of replacement Nvidia cards for the D620. 15 day wait list if yours fails.
Liberte, Egalite, Fraternite (TM)
Practically speaking, using an exponential distribution means this:
If you can expect a card to last (on average) one year when it's new, then, given that it's N months old, you can still expect it to last one year. An exponential distribution has no history.
It works surprisingly well.
Anyone familiar with Nvidia's hiring practices could've predicted a disaster of this magnitude. They have absolutely moronic screening quizzes (yes, quizzes)... canned technical questions with only one "right" answer. Throw original thinking out the window, they just want to hire drones to do their dirty work. You know what Nvidia, you get what you deserve. Maybe they had a majority of good employees at one time, but the way they treat their people, I imagine there's been alot of brain drain. It's a clear sign of a failing company when they rely on marketing & sales to hide shoddy engineering. You had agood run Nvidia, but the last stop is coming up. Get off while you can, the train derails up ahead.
So you didn't get the job then?
"NVidia is currently in the midst of a $200M recall of bad GPUs"
Last I checked, they reserved $200M on their financial sheets in case they needed to deal with the chips. I've heard nothing about an official recall? Only thing I can find is a lot of angry resellers who are demanding a recall.
Correct me if I'm wrong?
That's the whole point of this fiasco.
NVidia began to update their manufacturing process in the middle of the life cycle of several chips.
They switched to eutectic pads and a new underfill material (which has a lower glassification temperature).
They stuck with high-lead solder because the bumps are laid out very early on and changing them would be a major undertaking.
Basically, this causes shit to break down faster because the parts are now much more susceptible to thermal stress.
NVidia knew there were problems.
Laptop parts started failing at very high rates.
OEMs knew about the failures in November of 2007, or earlier.
NVidia blamed OEM designs creating thermal issues.
NVidia offered to foot half of the bill (replacements, handling customers, fixes, etc.) with DELL and HP.
DELL and HP jumped at the chance to have the massive bill cut in half (this kind of offer is unheard of).
NVidia's "fix" was to crank up the fans with a BIOS update.
OEMs found desktop parts were failing at alarming rates as well.
OEMs were forbidden from speaking out about the real issue (lumped in there with that "we'll pay half" deal).
OEMs find out that their designs do meet NVidia's recommended thermal and electrical constraints.
NVidia continues to sell existing bad cards that are on shelves, and makes no mention of any of this to customers or retailers.
NVidia switches away from high-lead solder, completing the updates to it's manufacturing process, and may now be pumping out good parts.
But these new parts have not had any power distribution / control changes to accommodate the new solder material.
These new parts will likely have higher-than-normal failure rates as well.
NVidia does NOT designate the new parts in any way on the box.
NVidia does not designate the new parts in any (official) way on the actual hardware.
News comes out (last week) that NVidia based chipsets (motherboard chipsets, e.g. nForce) are bad too.