WTF is up with NVIDIA?

[taltamir]

Oh please, nVidia gives out the specified wattage for its chips, and laptop/mobo engineers build the computer accordingly. (cooling design, fan speeds, etc...) If the chips run hotter than they're supposed to, how are Dell. HP, Apple supposed to know?

Quality control? testing?

Ofcourse it is a bit harsh to say it is THEIR fault for not noticing that nvidia estimated the heat generation lower then it really is. Assuming that is really what happened.

 

[semisonic9]

I bought a Dell Inspiron 9300 (which at the time was simply an XPS with a different GPU).

The only problems I ever had with that laptop were GPU-related. Two of them burned out on me in three years, which seems a bit much.

These days I simply don't rely on notebooks for gaming.

-S

 

[rjc]

Originally posted by: nRollo
Actually we don't know if it cost "at least $200m", w just know they set aside $200m. Could be more, could be less.

As far as the link goes, those stats are for laptop failure from all possible sources, not just GPU failure.

Yeah sorry, you were using percentages that were way too small overall, was just trying to give a better idea on notebook reliability. I remember reading when this first started up that normal gpus failure rate was around 2-3%(sorry have been trying to hunt for a reference) on this. Also nvidia originally set aside $25m for handling faulty products. So adding a $200m charge appears to be 8x the normal failure rate. ie if its 10-15% then normal failure is roughly 1-2%

I'd be very surprised if it's "10-15%" like Charles and Wolfgang speculate, I'd be sitting in airports and offices hearing "Damn! My laptop died!" then "Hey! Mine too!".

You are overrating nvidias market share in laptops. Have a look here:
http://www.jonpeddie.com/about...2008-gpu-shipments.php
Last table indicates that nvidia had a market share of 30% in the problem time.
ie 15% of 30% is 4.5% of total notebooks affected.

As an aside note also from the above figures the decline of the add in gpu in notebooks, this is likely to get worse with the rise of the netbooks.

This would have gotten a lot more press if 15/100 laptops were dying due to GPU alone, if you added in the other causes you might have 1/4 laptops defective. We buy 100s at work with NVIDIA chips- I've never heard of one dying on anyone.

Maybe turn off the air conditioning for awhile?

 

[nRollo]

Originally posted by: rjc

Originally posted by: nRollo
Actually we don't know if it cost "at least $200m", w just know they set aside $200m. Could be more, could be less.

As far as the link goes, those stats are for laptop failure from all possible sources, not just GPU failure.

Yeah sorry, you were using percentages that were way too small overall, was just trying to give a better idea on notebook reliability. I remember reading when this first started up that normal gpus failure rate was around 2-3%(sorry have been trying to hunt for a reference) on this. Also nvidia originally set aside $25m for handling faulty products. So adding a $200m charge appears to be 8x the normal failure rate. ie if its 10-15% then normal failure is roughly 1-2%

I'd be very surprised if it's "10-15%" like Charles and Wolfgang speculate, I'd be sitting in airports and offices hearing "Damn! My laptop died!" then "Hey! Mine too!".

You are overrating nvidias market share in laptops. Have a look here:
http://www.jonpeddie.com/about...2008-gpu-shipments.php
Last table indicates that nvidia had a market share of 30% in the problem time.
ie 15% of 30% is 4.5% of total notebooks affected.

As an aside note also from the above figures the decline of the add in gpu in notebooks, this is likely to get worse with the rise of the netbooks.

This would have gotten a lot more press if 15/100 laptops were dying due to GPU alone, if you added in the other causes you might have 1/4 laptops defective. We buy 100s at work with NVIDIA chips- I've never heard of one dying on anyone.

Maybe turn off the air conditioning for awhile?

Heh- you make some good points.
(and if you turn off the air, I'm goin' postal......)


I don't think we'll ever know what the true failure rates were, but it's good to see vendors extending warranties so people have recourse.

 

[cheezy321]

Bumping it for another update.

HP has offered free repairs for 38 models of its Pavillion Slimline range of desktop machines, which are having problems with Nvidia graphics units.

http://www.theregister.co.uk/2.../hp_slimline_problems/


This must be hurting NVIDIA pretty bad....

 

[ViRGE]

Originally posted by: cheezy321
Bumping it for another update.

HP has offered free repairs for 38 models of its Pavillion Slimline range of desktop machines, which are having problems with Nvidia graphics units.

http://www.theregister.co.uk/2.../hp_slimline_problems/


This must be hurting NVIDIA pretty bad....

Repost

 

[OCChronic]

First of all, whether laptops are meant for gaming or not is a moot point and not the heart of the issue. Also, modifying the chipset is not something you should have to do nor is it something you should do because it will void the warranty if your chipset fails. This is exactly what the OEM's want because they are off the hook because you void the warranty when doing this.

Better quality control and better testing in real world environments is what should hvae happened by both the chipset maker(nVidia) and the OEM builder(Dell, HP, Compaq, etc.). Both segments of the market dropped the ball completely and let an inferior component, that is prone to early failure, slip into their products and cause countless hardships and cost consumers possibly millions of dollars in data loss and inconvenience.

For example, if Dell makes a desktop machine and the PSU doesn't have proper safety features and catches fire and burns your house down, no-one is going to say, 'Hey switch out the PSU with an Antec to solve that problem'. It's rediculous to suggest the average user, who knows nothing about the guts of their laptop/desktop to modify it in any way for better cooling or claim that everything is alright as long as you don't do any real gaming on your laptop/desktop using this chipset. Besides, gaming would only raise the temps of their GPU by a few degrees Celcius and shouldn't matter. That's why some poeple bought their system, using this chipset because it boasted better gaming performance than other alternatives and they wanted a cheap light gamer system, whether laptop or desktop.

I love nVidia and they have solid products that usually outperform AMD/ATI, produce less heat & use less power but they definitely dropped the ball on this one and the OEM's didn't research the problem and test the chipsets enough before making the decision to use them.

 

[jiffylube1024]

Originally posted by: Zebo
Louissssssssssssssssssss I saw that and also read the real original quote where Mr Jensen seemed to indicate thermal cracking was also attributable to poor software and encloser design as well. I'm not saying Apple is responsible. What I'm saying is this claim about ALL nV card are defective is BS. I've owned every gen since TNT and never had an issue. How could they stay in business with a huge failure rate let alone lead the industry with all their cards being defective?[

The power consumption and heat output from GPU's from even two years ago are way different from the GPU's of today. Remember, the PCI-express power connector was added to deliver extra power to the GPU over a 4-pin molex, and the PCI-express 16x slot also can deliver a lot more power than the AGP slot.
------------


This problem with nvidia GPU's that may or may not be caused by the substrate material, etc. very likely would not have even surfaced if the GPU's weren't running so hot, and taking so much power these days.

But the most important confounding factor is the concentration of heat in a tiny area is a problem that die shrinks exacerbate. The kinds of problems that you get on 65nm/55nm GPU's (of today) are very different from ones on .35um and .25um (of 5-10 years ago). This is sort of similar to the problems Intel had when going from 130nm Northwood to 90nm Prescott P4's and we all got to learn about "power leakage" in CPU's.

---------------

The claim that "all Nvidia GPU's are affected" by this problem should really be clarified to "all 65nm Nvidia GPU's", and that just beacause a GPU is affected by this problem doesn't mean that it will die a premature death, just that its likelihood of early death is increased and that heat is a problem.

Heck, I've always been alarmed by the crazy temperatures of ATI and Nvidia GPU's, since the ATI X800 Pro / Nvidia GeForce 5 days, and I routinely switch to aftermarket cooling to drop my temperatures from 70-80C at load to mid 40's.

 

[OCChronic]

Originally posted by: jiffylube1024

But the most important confounding factor is the concentration of heat in a tiny area is a problem that die shrinks exacerbate. The kinds of problems that you get on 65nm/55nm GPU's (of today) are very different from ones on .35um and .25um (of 5-10 years ago). This is sort of similar to the problems Intel had when going from 130nm Northwood to 90nm Prescott P4's and we all got to learn about "power leakage" in CPU's.

The claim that "all Nvidia GPU's are affected" by this problem should really be clarified to "all 65nm Nvidia GPU's", and that just beacause a GPU is affected by this problem doesn't mean that it will die a premature death, just that its likelihood of early death is increased and that heat is a problem.

Heck, I've always been alarmed by the crazy temperatures of ATI and Nvidia GPU's, since the ATI X800 Pro / Nvidia GeForce 5 days, and I routinely switch to aftermarket cooling to drop my temperatures from 70-80C at load to mid 40's.

You're partly right. Actually, to correct your thinking, I have to point out that a die shrink generally yields a chip that consumes less power and produces less heat, clock for clock. The Northwoods were hot chips because Intel created a long pipeline to increase clock speeds but they had to pump more voltage throught them to obtain those high clocks which in turn created more heat.

ATI and nVidia do pump a lot of data through their GPU's and like the northwoods are watt hungry and therefore heat up like crazy. 85c is not unheard of for these GPU's but the chipset integrated graphics are a bit different and shouldn't get as hot. Like I said, both nVidia and the OEM's who chose to go with these chipsets dropped the ball because they did not do enough research and quality control to ensure these chips were rugged enough for real world use.

 

[sxr7171]

Originally posted by: amdskip
Laptop does not = gaming, if you want to game get a desktop

Thank you for helping put some sense into people. Seriously, you have people arguing about graphics cards in ultraportables of all things. To me, on a laptop, if I get an image on the screen the graphics card is doing its job. Just make the damn thing Office and my browser properly and keep it as small and light as possible. I think people are just swayed by marketing that caters to spec whores. They think they know everything because they can read specs off a spec sheet and determine which number is higher.

 

[sxr7171]

Originally posted by: SarcasticDwarf

Originally posted by: amdskip
Laptop does not = gaming, if you want to game get a desktop

The XPS M series (M1330, M1530, M1730) are laptops designed for gaming. And while they do not replace a full rig they are a good temporary substitute.

You make me laugh. Gaming laptop = oxymoron. Marketing always gets the best pf people.

 

[sxr7171]

Originally posted by: apoppin

Originally posted by: Zebo
Yeah right .. So one vendor, Apple , a megalomaniac personality BTW, wants to push their problems off on a vendor. Who happens to sell the most cards by far and with very deep pockets. Sure when you're in every other lappy that doesnt have I965 you're bound to come up with issues. Laptops were not made for these hot beasts they make today.

I would also question before jumping to conclusions is who designed the board for apple's (especially apple knowing little Stevie) and dells laptops? Then my next question is who commissioned construction of said PCBs? The part nV delievers, the chip - has no problems and they oversee it's design to delivery. I noticed the article left that part out.

Zebo .. it is way beyond Apple; it is widespread. My partner Dave McOwen has one of these overheating defective chipsets in a Toshiba notebook [i believe] - and he is a nvidia fanboy to the max

Nvidia screwed up; they *admitted* it and took the blame - $200 million dollars worth [so far] even though TSMC may also be partly to blame

read the whole article or google it

We previously reported that the GPU packaging problem, widely believed to be caused by a cracking solder bumps, may be much more extensive than was admitted by Nvidia. High lead was used in more than 70 million Nvidia GPUs overall and more than 15 million mobile GPUs. However, the problem of fatigue cracking appears to be limited to a relatively low percentage of notebook GPUs ? and only Nvidia knows how high that percentage is. Nvidia recently put aside $200 million to cover notebook repairs, which suggests that somewhere between 500,000 and 1 million notebooks were affected ? if the general estimate of $200-$300 repair cost estimate is right.

Two lawsuits (here and here) alleging that Nvidia has not been truthful about the packaging issue have been filed so far. Both lawsuits are seeking class action status.

Where have you been?

[/quote]

He's been smoking with Rollo apparently.

 

[sxr7171]

Originally posted by: Spicedaddy

Originally posted by: Zebo

Originally posted by: LOUISSSSS

Originally posted by: Zebo
Yeah right .. So one vendor, Apple , a megalomaniac personality BTW, wants to push their problems off on a vendor. Who happens to sell the most cards by far and with very deep pockets. Sure when you're in every other lappy that doesnt have I965 you're bound to come up with issues. Laptops were not made for these hot beasts they make today.

I would also question before jumping to conclusions is who designed the board for apple's (especially apple knowing little Stevie) and dells laptops? Then my next question is who commissioned construction of said PCBs? The part nV delievers, the chip - has no problems and they oversee it's design to delivery. I noticed the article left that part out.

what do u mean? if a laptop was sold in the market with that specific gpu, it was clearly made for it. i'm sure that the design of the pcb for the gpu was made to meet or exceed nvidia's specifications.
This is clearly a very large problem with the nvidia mobile gpu's (are the problems limited only to mobile gpu's?)

I mean nV sure didn't design the constricted case it sets in. Was dells/apples designs ok with nV or not? Or were they even asked? I mean typically laptop makers design thier own mobos in conjunction with PCB makers. What part, if any, did nV play? These are all important questions I have not seen answered in links I've read. If I throw my 280GTX in a shuttle and cut off the fan I'm sure it will burn up too. Likewise so will a 4870 burn up. It's only a matter of scale.

Oh please, nVidia gives out the specified wattage for its chips, and laptop/mobo engineers build the computer accordingly. (cooling design, fan speeds, etc...) If the chips run hotter than they're supposed to, how are Dell. HP, Apple supposed to know?

The fact that it's happening with a few vendors means the problem is from a bad batch of nVidia chips.

Seriously. No major computer manufacturer is stupid enough to design a PCB without considering the thermal data they were given. They are not in the business of having to fix millions of laptops over a bad PCB design. They take care about such things. Now Microsoft is another thing altogether....

 

[Keysplayr]

Originally posted by: sxr7171


Seriously. No major computer manufacturer is stupid enough to design a PCB without considering the thermal data they were given. They are not in the business of having to fix millions of laptops over a bad PCB design. They take care about such things. Now Microsoft is another thing altogether....

Yeah ok. Here's a quick search.

SONY major laptop recall (500,000 units) due to faulty wiring (not related to the exploding batteries)

DELL A/C adapter fire hazard recall (990,000 units)

Because they take care about such things. :roll: