Companies are not allowed to lie to their investors. Part of the deal of being a public company is that all communication is communication with their investors. If they told their investors that they were selling GPUs to gamers, while in fact selling most of them to miners, they violated the law. The principle of it being that said investors might have had a different outlook on the company depending on who they were selling to.
Besides, there is a difference. Because whenever crypto collapses (and it has twice already in the past), a lot of the existing mining GPUs end up on the secondary market. If this happens now, it will have a material impact on the prices of GPUs going forwards.
> they told their investors that they were selling GPUs to gamers, while in fact selling most of them to miners
I don't think nvidia even did this, they marketed to gamers and reported the correct number of sales. What people do with the card once its sold is not up to nvidia its not information that an investor can reasonable assume nvidia will report on.
If I invest in a steel mine they market their steel to car manufactures and report the correct number of sales. If they also happen to sell to gun manufactures I can't then sue said company for not reporting that.
On the other hand, NVIDIA does not have legal authority to control where its GPUs actually end up being used (nasty things like drivers shutting down when detecting virtualizations aside). How can anyone prove or disprove that a specific percentage of GPUs end up in a specific market, as long as NVIDIA can prove they have kept their usual delivery targets towards wholesale distributors?
That is irrelevant. The SEC also didn't go and check what people were using their NVIDIA GPUs for.
The problem is that NVIDIA leadership knew/believed that a proportion X of their GPUs were being bought for crypto mining (thus tying demand with crypto prices), while they reported to investors a different number, Y, with Y < X. This is material information that they simply lied about, which is illegal for a publicly traded company. In fact, the lawsuit could have happened even if Y > X.
If NVIDIA had reported the same numbers they did, and people had been in fact using them at the same rate that they did, but NVIDIA hadn't known about it, then nothing illegal would have happened. If NVIDIA had later found out and disclosed this to investors, again no lawsuit (though perhaps share prices would have changed, one way or another).
This is a good question. I remember having to give my identity before walking out of a store with a computer with a GPU in it, and if someone really wanted to buy multiple GPUs for mining that wouldn't stop them.
"NVIDIA launched a product line of cryptomining processors, known as “CMP,” which the company marketed to large cryptomining operations ... Based on known CMP sales, the company identified cryptomining as a significant element of the OEM GPU sales"
"NVIDIA also received information indicating that cryptomining was a significant factor in year-over-year growth in NVIDIA’s Gaming GPUs revenue. Some of the company’s sales personnel, in particular in China, reported what they believed to be significant increases in demand for Gaming GPUs as a result of cryptomining"
These were the 2 most concrete points to me - management was aware miners drove up GPU prices for gamers and saw it as another big market to capture, and sales attributed increased demand to miners. The rest of it basically says crypto grew, nvidia sales grew, and nvidia thought sales grwe because of crypto.
Honestly, I'm not sure it's possible even knowing these factors to know who really bought the GPUs or where they really ended up, but idk all the SEC probably cares about is nvidia not disclosing material seeming information.
No; there's been many tests. The mistake seems to come from thinking GPUs "wear out" like cars. The cooler might, but the silicon is fine.
Some even suspect that crypto mining is better for the card than gaming as it avoids the constant ramp-up/ramp-down of the clock/activity by keeping it pegged at 100%. In turn, that would lower thermal stress.
Actually there's reasonable evidence to believe this isn't true. This paper from google: https://sigops.org/s/conferences/hotos/2021/papers/hotos21-s... outlines specific scenarios where CPUs fail over time. Given the evidence that these are silicon defects that are actually worsening over time, there's no reason to imagine these failures don't extend to GPUs as well.
The difference in data here is obviously scale, google has -way- more CPUs than GPUs so the absolute counts of failures will be different.
I'll concede that silicon can wear out over time; it's impossible for it to be immune from happening, and I wasn't speaking in absolutes. But as you mention, it's one of scale. I'm curious how likely a second hand GPU from a cryptominer is affected.
So that's actually one of the awful implications of this paper. It's probably actually happening at a rate higher than would be noticed by humans.
If a given piece of silicon is hosing up a GEMM (matrix multiply), in graphics scenarios this may be invisible to the human eye as it could potentially just introduce artifacts in a scene rendering that could be entirely ephemeral to the frame.
In the case of crypto mining though, it's completely possible (probable?) that there are GPUs that can't possibly ever calculate a proper SHA3 hash (see the paper on AES instructions that fail in symmetric ways).
Yes, because hard drives, like cars, have moving parts that can wear out.[a] The silicon of a GPU doesn't have moving parts and is therefore more resilient.
[a]: It's (hopefully) common knowledge in the tech community that purchasing used hard drives is a Bad Idea(TM)
Yes, but for different reasons. The grandparent is probably incorrect though, there is emerging evidence that silicon is actually changing/failing over time. See this paper from google on their CPU cores where they have practical evidence of this occuring: https://sigops.org/s/conferences/hotos/2021/papers/hotos21-s...
If their data is correct, it should follow that these exact issues will happen on the small transistor process GPUs as well.
It's not emerging, electromigration been a core constraint in semiconductor design for 60 odd years. E.g.,:
Blech, I. A. and Sello, H. (1966). The Failure of Thin Alluminum Current-Carrying Strips on Oxidized Silicon. In Proc. Symposium on PoF in Electronics, 496-505.
Totally untrue. I've run many many tens of thousands of GPUs for years, in the worst of conditions (directly exposed to seasonal temperatures and environment)... they keep on ticking.
Power supplies on the other hand... those don't seem to last as long, but we open those up and they are hand soldered by someone in China.
Actually it can be the opposite - if the miner is using it properly and not overheating them regularly they can be in good shape. If they run at a high percentage but the temperature stays constant then theres not a lot of wear going on vs suddenly heating and cooling all the time.
I imagine the bearings in the fan might have more wear, however, but that might be more easily fixed.
Besides, there is a difference. Because whenever crypto collapses (and it has twice already in the past), a lot of the existing mining GPUs end up on the secondary market. If this happens now, it will have a material impact on the prices of GPUs going forwards.