NVIDIA shows signs ... [2008 - 2017]

[AlphaWolf]

That sounds exactly like what happens in a laptop, doesn't it ?
There are no reference heatsink/cooler for notebooks like there are in desktop chips, at least to my knowledge (underfill material issue aside).

Microsoft, long before issuing the 1.1B repair charge notice, repeatedly kept sending costumers faulty consoles -some guys got seven (!)- before they admitted anything, going as far as having top-level executives publicly and categorically denying the abnormally high RMA rates.

No, its not the same. Ati isn't responsible for producing the chips for the 360, they developed the IP, not the product. It's similar from a perspective of MS and Nvidia, in the way that MS handled the issue is similar to the way Nvidia is handling the issue, but I haven't heard MS blaming any of their suppliers for the problems yet.

 

[INKster]

but I haven't heard MS blaming any of their suppliers for the problems yet.

MS itself didn't, but an insider did:

The main design flaw was the excessive heat on the GPU warping the mother board around it. This would stress the solder joints on the GPU and any bad joints would then fail in early life.

http://www.8bitjoystick.com/archive...about_xbox_360_red_ring_of_death_failures.php

 

[Sxotty]

I'm not so sure I'd be quite so sympathetic to the multi-billion dollar company that sold me a known faulty product that fails just outside of warranty and leaves me to pick up the cost and inconvenience of a replacement or repair.

If Charlie's other claims are true about this being a known problem for the last year or so, and Nvidia continuing to ship out faulty chips to get rid of them, I'd be pretty pissed off.

Sure, AMD had TLB, but that was spotted and owned up to within a couple of months. There was a software fix and at no point did your laptop or a major componet die. In fact, didn't AMD stop shipping to retail to make sure?

It's all pretty different from the shennanigans that Nvidia's been pulling about all of this.

That really isn't accurate b/c their "software fix" decreased performance. Nvidia could easily implement a "software fix" as well by that standard. Simply lower the clocks massively.

 

[Speccy]

MS itself didn't, but an insider did:

Subsequent reports indicate that the issue stemmed from the MS design implementation and the resolution was done by going back to the ASIC vendor to implement again:

http://www.eetimes.com/news/semi/sh...d=SEDOVWZF43Y14QSNDLSCKHA?articleID=208403010

The Xbox 360 recall a year ago happened because "Microsoft wanted to avoid an ASIC vendor," said Lewis. Microsoft designed the graphic chip on its own, cut a traditional ASIC vendor out of the process and went straight to Taiwan Semiconductor Manufacturing Co. Ltd., he explained.

But in the end, by going cheap--hoping to save tens of millions of dollars in ASIC design costs, Microsoft ended up paying more than $1 billion for its Xbox 360 recall.

To fix the problem, Microsoft went back to an unnamed ASIC vendor based in the United States and redesigned the chip, Lewis added. (Based on a previous report, the ASIC vendor is most likely the former ATI Technologies, now part of AMD.)

Indicates that said ASIC vendor knew about these risks hence why MS went back to them.

 

[XMAN26]

I never said "fallling apart". The connections between the chip and the pads on the substrate are breaking due to thermal stress and migration.

How do you explain that Nvidia is replacing all underfill material in all it's chips, and switching to eutedic balls, even for chips that are effectively reaching the end of their lives anyway? It's very expensive to rebuild and requalify this kind of thing, it's not something you just do for a laugh, or drop 200 milliion on and hurt your stock price on unless you really, really have to.

Read the article, because I can't believe that you have done so if you have to keep asking these questions. Charlie has got the spec of the old underfill material, and it's rated only up to 60 degrees, and at 80 degrees it's a hundred times weaker. Nvidia chips run hotter than that at the balls and pads, which are effectively heatsinks into the chips themselves. This underfill failure is exactly what Nvidia pinpointed themselves when they finally admitted the problem, but you can see the underfill isn't rated for the very high temps seen on the balls/pads between the chip and it's substrate. The underfill works within it's spec. Nvidia is using it way outside it's specification.

The failures are caused by the materials and the way they interact with the heat stressing caused by the design and nature of these chips. Some will last longer than others because of the environment they are in, how they are cooled, how they are used. Others will fail more quickly for the same reasons, and laptops are currently the worst case scenario for failures. They are not the only case scenario though.

my take on the whole issue,
1. its true and it was an issue, they have changed the substrate that was used in bonding

2. Nv is the blame as much as the assembler in this mess. When you design a BIOS to run your fan at 20-40% even under load, is it no wonder you end up with heat related issues. I BIOS flash all my Nv based products to 100% fan, 100% of the time and I have not had 1 fail yet due to heat related issues.(1 8800GTS 640OC and also BIOS OC'd, 2x XFX 9600GT XXX, BFG GTX260OC, eVGA and XFX OC 8600GTs) Had them all over 1 yr now and not a fail yet.

3. All laptop companies have their role in this as well, see #2 and add in trying to make things as small as possible and light weight. Gaming cards are not something you should do this to, especially if they are going into a lappy.

Part 3 has more baring on this discussion than anything.

 

[aaronspink]

That sounds exactly like what happens in a laptop, doesn't it ?
There are no reference heatsink/cooler for notebooks like there are in desktop chips, at least to my knowledge (underfill material issue aside).

Microsoft, long before issuing the 1.1B repair charge notice, repeatedly kept sending costumers faulty consoles -some guys got seven (!)- before they admitted anything, going as far as having top-level executives publicly and categorically denying the abnormally high RMA rates.

Nvidia defines the specification for the thermal performance of the cooling for the gpu. AFAIK, the gpus are generally being used with nvidia reference cooling solutions as well.

 

[Bouncing Zabaglione Bros.]

That really isn't accurate b/c their "software fix" decreased performance. Nvidia could easily implement a "software fix" as well by that standard. Simply lower the clocks massively.

And like the AMD fix, would render Nvidia GPUs very uncompetitive.

 

[Bouncing Zabaglione Bros.]

2. Nv is the blame as much as the assembler in this mess. When you design a BIOS to run your fan at 20-40% even under load, is it no wonder you end up with heat related issues. I BIOS flash all my Nv based products to 100% fan, 100% of the time and I have not had 1 fail yet due to heat related issues.(1 8800GTS 640OC and also BIOS OC'd, 2x XFX 9600GT XXX, BFG GTX260OC, eVGA and XFX OC 8600GTs) Had them all over 1 yr now and not a fail yet.

Most people would consider that unacceptable. I would never buy a product where I had to run the fan at 100 percent all the time for fear of complete failure. Just too noisy for a home PC environment.

 

[Geo]

Is it just me or did it get a little warm in here today?

Members will please recall that B3D is one big happy family and will post accordingly.

 

[XMAN26]

Most people would consider that unacceptable. I would never buy a product where I had to run the fan at 100 percent all the time for fear of complete failure. Just too noisy for a home PC environment.

For me its a none issue, fan noice doesn't bother me at all, I wear headphones and dont even hear the fans running when at 100%. Other, yes it might be, but again, it is an issue Nv made. Who in their right mind would think 80+c degrees for load temps would be perfectly ok for a GPU and have a fan spinning at anything less than 100% is plain loco. My cards never go over 70c with the fans set to 100%, and that is with OCs above the factory ones. And I do it not for fear of failure, but because I prefer my temps to stay as low as possible.

the 8800GTS640OC(without my bios mod) would hit 85c, with it, never went over 70 and that is with my BIOS mod OC aswell, the 9600GTs by XFX would hit 73, with the mod, 62, the 8600s would also approach 70 before the mod, now they barely ever hit 60c. My 260 has yet to even go over 55c since doing it.

 

[Sxotty]

I honestly did not see much sniping or personal attacks, but I may have overlooked them.

 

[Skrying]

For me its a none issue, fan noice doesn't bother me at all, I wear headphones and dont even hear the fans running when at 100%. Other, yes it might be, but again, it is an issue Nv made. Who in their right mind would think 80+c degrees for load temps would be perfectly ok for a GPU and have a fan spinning at anything less than 100% is plain loco. My cards never go over 70c with the fans set to 100%, and that is with OCs above the factory ones. And I do it not for fear of failure, but because I prefer my temps to stay as low as possible.

the 8800GTS640OC(without my bios mod) would hit 85c, with it, never went over 70 and that is with my BIOS mod OC aswell, the 9600GTs by XFX would hit 73, with the mod, 62, the 8600s would also approach 70 before the mod, now they barely ever hit 60c. My 260 has yet to even go over 55c since doing it.

You're well into the minority. I wouldn't dare take my video card fan over 50%, let alone to 100%. Even with headphones you're going to hear that, even if they're closed. I would seriously question your hearing if you can't hear that, or you're listening to things WAY to loud. Also, the load temperature at max should be a function of what the GPU can handle. If it's fine up in those ranges (which it should be) then there is little reason for concern. Most people value silence to some degree.

 

[XMAN26]

You're well into the minority. I wouldn't dare take my video card fan over 50%, let alone to 100%. Even with headphones you're going to hear that, even if they're closed. I would seriously question your hearing if you can't hear that, or you're listening to things WAY to loud. Also, the load temperature at max should be a function of what the GPU can handle. If it's fine up in those ranges (which it should be) then there is little reason for concern. Most people value silence to some degree.

Nope sorry, dont and the volume isn't that high, now if i put the case on my desktop and remove the side panel, yes I can hear it, but with the panel on and it sitting on the floor about 3-4ft away from ears, I dont hear it at all wiht the headphones on. I got good ones too, they kill almost 100% of outside noise. Hell, my wife has to yell or tap my shoulder when I'm gaming because otherwise i wont hear her. Was a nice 50 buy from CUSA before they died.

And yes, I am in the minority and I have no problem with that.

 

[Florin]

Is it just me or did it get a little warm in here today?

Members will please recall that B3D is one big happy family and will post accordingly.

I think this means Geo needs a hug. Scrum time everyone!

 

[Florin]

2. Nv is the blame as much as the assembler in this mess. When you design a BIOS to run your fan at 20-40% even under load, is it no wonder you end up with heat related issues. I BIOS flash all my Nv based products to 100% fan, 100% of the time and I have not had 1 fail yet due to heat related issues.(1 8800GTS 640OC and also BIOS OC'd, 2x XFX 9600GT XXX, BFG GTX260OC, eVGA and XFX OC 8600GTs) Had them all over 1 yr now and not a fail yet.

Whatever works for you It's not strictly necessary tho, I've never had an Nvidia card fail period (except for the defective-by-design 6800LE). But then I don't overclock either.

 

[stevem]

Is the Gt200 ASIC likely affected by this combination of construction, die layout, & power draw, too?

I've never had an Nvidia card fail period (except for the defective-by-design 6800LE).

Simple capacitor fix for the black screen issue.

 

[XMAN26]

Whatever works for you It's not strictly necessary tho, I've never had an Nvidia card fail period (except for the defective-by-design 6800LE). But then I don't overclock either.

Neither have I, but then again, I'd prefer to cooler running GPUs than really hot ones. My main rig with the 260 sits in a small room with one AC vent so it gets quite warm. The longer it takees to warm up the room, the better.

 

[I.S.T.]

I'm not sure i'm comfortable with your personal attack against me. I certainly didn't do it to you.

BTW, Xbox 360 hardware is sold at a loss for Microsoft (which can afford to cover the Xbox/Zune division continuing losses since 2001 with the Windows/Office software licensing revenues), and most, if not all of the RRoD's stemmed from overheating ATI GPU's (it certainly wasn't the IBM CPU).
Nvidia just hands over the GPU for implementation into the OEM motherboards, and expects profits for it, like its reasonable for any company hoping to strive.
I remember seeing the Xbox 360 and PS3 motherboards side by side on a website back in 2005, and commented to a friend just how littered with cheap electrolytic capacitors was the former, compared to the neat and clean PS3 layout.

was sold for a loss. It isn't anymore.

 

[aaronspink]

You're well into the minority. I wouldn't dare take my video card fan over 50%, let alone to 100%. Even with headphones you're going to hear that, even if they're closed. I would seriously question your hearing if you can't hear that, or you're listening to things WAY to loud. Also, the load temperature at max should be a function of what the GPU can handle. If it's fine up in those ranges (which it should be) then there is little reason for concern. Most people value silence to some degree.

I think we can all agree that most of the video card cooling solutions leave much to be desired.

I've long since given up on stock cooling solutions, they are incredibly compromised, quite honestly, I wish I could buy the cards without a cooling solution to save money and aggravation, since the first thing I'm going to do is put a decent cooling solution on the thing. In other words, my video card runs <50c at load with noise levels in many cases lower than the stock cooling at min voltage.

 

[aaronspink]

Is the Gt200 ASIC likely affected by this combination of construction, die layout, & power draw, too?

The current GT200 likely catches a break due to the thermal flux being so spread out across the die. But a lot of it depends on the die temps for it as well. Most designs aren't designed to run much above 80c Tj continually. And sadly thats where most of the GPUs run. Assuming the cooling solution is decent, one should never see this problem.