This makes a lot of sense for NVIDIA. They have the expertise, the money, the scale, and the experience already. They can probably do it cheaper than any startup and then either pass on that savings or make more profit.
Has Nvidia ever passed on savings? It seems to be the opposite to how they operate.
Turing and ampere as a whole were already below the price increase curve due to the trailing node. Ampere certainly was a massive attempt at passing the savings through - 3060 ti and 3080 were both really aggressively priced and were great overall products. Notoriously, 3060 ti was so aggressively priced that partners didn’t even want to make it at msrp…
And Turing is another example of the “AMD is subject to the same overall industry price trends/not able to drastically beat nvidia pricing either” too - sure, AMD leapt ahead to 7nm early with 5700xt… and this was more expensive than Nvidia’s solution with Turing, such that AMD was not able to undercut Nvidia’s pricing. Those cost increases have been eating up the density gains for a long time now.
Nvidia is really actually doing the things that keep the costs down, and ampere and Turing indeed bent the cost curve below where it would otherwise have been. It’s obviously not going to stay fixed, the era of $100 and $150 gaming gpus being serious contenders has obviously long since come to a close, and now it’s happening for $200-300 gpus too, $500-600 is where a $300 gpu was 10 years ago.
People just don’t have any way to objectively determine what “true” prices should be, and refuse to believe the true price is rising because it offends their “gamer” worldview. It has to be gouging, otherwise I don’t get a new Sony GameStation X every 2 years at the exact same price, and that can’t be possible.
https://www.pcgamer.com/amd-moores-law-aint-dead-its-just-a-...
You've got to be joking. The 3000 series cards weren't any kind of "pass along savings", and this generation after it have pricing best called "taking the piss". Widely regarded as "price gouging".
You're probably the very first person on earth to try telling people those cards are cheap (with a straight face). I don't think you're going to be able to convince anyone that you're right.
I've said it before and it's true today: a lot of people don't think of themselves as fanboys, they just constantly say all the same things fanboys say and think all the things fanboys think. The ayymd discourse permeates the online discussions of all these things such that people can no longer determine what's biased and what's not.
Moreover, as I've said before, a lot of people's sense on pricing today is just miscalibrated and often they're outright hallucinating on past prices. GTX 970 was the cheaper x70 ever released - the segment has bounced between $350 and $400 since it was introduced, and $349 in 2010 was a lot more money than it is now. And people don't have a good sense of just how much costs have grown and are growing - most people don't even bother accounting for CPI in these discussions.
https://en.wikipedia.org/wiki/List_of_Nvidia_graphics_proces...
GTX 670: $533 in 2023 dollars (cutdown of 294mm2 die)
I don't really understand how they can make the economics of it work. which worries me, especially as their AI work is just printing money. I wonder if it's just a matter of time before they abandon the gaming market.
Seems potentially reasonable if that's amortised over a few years.
The other cloud gaming providers need to do the same thing, and they either cost more and/or have worse hardware and/or have publisher relationships. Xbox and PS streaming use actual consoles and are much less powerful, for example, Shadow costs a lot more for worse hardware, Luna works with publishers, Stadia just gave up before ever really trying (lol), etc.
Maybe in Nvidia's case, at least they can get the GPUs at cost, and maybe use them for non gaming async workloads during off peak times? I dunno. I just hope they keep GFN going; it's completely replaced my desktop PC.
What innovation are they going to bring to the market? 3D XPoint combined with AI engine chiplets for petabyte scale AI models? That wouldn't cost more than a hundred billion.
The entire Abu Dhabi Investment Authority sovereign wealth fund isn't even 1 trillion.
7 belt and road initiatives?
The simple explanation is that our media is mind numbing trash that has little to do with reality other than we wouldn't be talking about this if it wasn't for the fake 7 trillion dollar number.
There's a company out there actually doing this: https://corticallabs.com/
There are outstanding problems, particularly I've found it very crash prone on a consumer desktop and wouldn't recommend an AMD card for research compute tasks where you are also running an X server using the same card. But there aren't $30 billion opportunities for custom chips on the consumer desktop right now - I'm guessing these will be for SaaS businesses where AMD are focusing. IE, it won't matter that they can't X.org and multiply matrices at the same time because servers won't use the cards for graphics.
The hard part is getting a culture that gives a damn about developing software that works and designing the hardware to support the features that the software needs.
AMD has not figured out how to run both graphics and compute on the same GPU. There can be many reasons for that, but honestly it is probably because they either don't have the necessary virtualization hardware or because two different drivers are conflicting with one another.
NVIDIA isn't missing the mark on the programming model and toolkit framework (PTX and forward/backward compat) either. They have a good, lean gpu design with a lot of features and a good programming model and ecosystem etc.
You're right, it's not just the matrix math, that's not rocket science, but there's a ton of little glue code around it. And you need something GPU-like for that anyway, plus a bunch of scheduler and shader-execution-reordering stuff for your tensor threads and glue code, etc. You end up with something broadly similar to a GPU anyway.
It's the ProgPOW theorem, right? That there is not some major gain to be squeezed by implementing a smaller/different machine on the instruction set. That GPUs are relatively close to some kind of computational optimum for parallel workloads (in terms of programmability/flexibility and performance).
NVIDIA's model isn't far off the global optimum imo, it's certainly in a great local minimum, and that's really true of a lot of their designs these days. It is always a little wild how everyone trivializes the idea that AMD/etc are going to catch up with some 80% solution in RT or tensor etc... like just maybe NVIDIA did the math and figured out what they think a reasonable ray performance level is, and how much they'd need to upscale, and what parts of the pipeline make sense to have accelerated by units vs emulated on shaders/etc, and there's not some massive gain to be squeezed by just putting a handful of devs on a project for a year?
Same thing for prices too. Everyone wants to assume that AMD is just choosing to follow them in gouging or whatever. The null hypothesis is that both nvidia and AMD are subject to the same industry cost trends and can’t actually do significantly better (not like 2x perf/$ or whatever), and that nvidia is in some kind of reasonable price structure after all. People are going to find that a lot of electronics prices are going to go up in the coming years. There’s no more 1600AF for $85 or 3600 for $160 either, or Radeon 7850 for $150 etc.
Not talking about original 4080 pricing etc but actually 4070 and 4060 are fairly reasonable products, and 4070 quickly fell even further below msrp. 7800xt and 7900xt and 7600xt are all fine as well. That’s about what the price increases have been since the last leasing-edge products.
https://rocm.docs.amd.com/projects/install-on-linux/en/docs-...
Although, as mentioned, works under a cycloptic vision where the card is only doing compute tasks. I'd be interested to know if even supported cards can multitask GPU and pure compute tasks without crashing because it looks like it might be a design issue. Hard to tell with driver corruption. Maybe the testing catches that on supported cards; who knows.
"“It’s not based on any particular data point,” a Treasury spokeswoman told Forbes.com Tuesday. “We just wanted to choose a really large number...”" - https://archive.thinkprogress.org/treasury-explains-how-it-c...
[0] https://www.britannica.com/event/World-War-II/Human-and-mate...
[1] https://moneywise.com/life/lifestyle/financial-facts-about-w...
[2] https://www.whatitcosts.com/world-war-ii-cost-united-states-...
Chip making is expensive, but what do you do with 7T? Buy out every engineer on the planet with half of it, and have them work on problems? Does Altman think he’s Oppenheimer?!
Every bespoke chip will be much more expensive - and profitable - than the generic units that were not bought.
Custom chips are likely made in technology nodes step behind They are cheaper to manufacture. Nvidia's H200 and Apple M2 are so profitable that they get the latest technology nodes.