Nvidia CEO says ARM smartphone chips with GeForce GPUs are not coming any time soon, AMD and Samsung can get the edge with the upcoming RDNA2-powered Exynos SoC
AMD dropped quite a few bombs during this year’s Computex keynote, including the teaser for the upcoming Samsung Exynos smartphone SoC that integrates a Radeon GPU with ray tracing capabilities. As far as graphics processing goes, smartphone SoCs have come a long way in just a few years thanks to the advancements pioneered by ARM. PC and consoles still have the edge, but if we starting seeing more chips like those coming from Samsung and AMD, the gap may be reduced significantly. Interestingly enough, ARM is in the process of being acquired by Nvidia, so the AMD-powered Exynos chips might get some serious competition, but not too soon, as suggested by Nvidia’s CEO in a Q&A session at Computex.
Jensen Huang stated that implementing a GeForce GPU with ray tracing features on a smartphone chip is not really a good idea for now:
Ray tracing games are quite large, to be honest. The data set is quite large, and there'll be a time for it. When the time is right we might consider it.
Huang is probably thinking that the GPU technology still needs to undergo a few more miniaturization cycles before combining it with smartphone chips. Samsung and AMD already seem to have 5 nm SoCs ready, so it is easier for them to do the implementation. Additionally, the ARM acquisition is not yet completed, so Nvidia may also need to wait at least until the first half of 2022 before further exploring this possibility. In any case, Huang believes that Nvidia’s current best solution for Android devices is the GeForce Now cloud gaming service that already caters to 10 million players across the globe and has a roster of 1000 games:
That's how we would like to reach Android devices, Chrome devices, iOS devices, Mac OS devices, Linux devices, all kinds of devices, whether it's on TV, mobile device, or PC. I think that for us, right now, that's the best strategy.
Speaking of miniaturization processes, the media also inquired about possible Nvidia investments in proprietary fabrication nodes. As of now, Nvidia is fabless and relies on foundries like Samsung and TSMC. The ARM acquisition could in time bring enough funds for proprietary production nodes, but Huang is not very keen on such endeavors:
There's a difference between a kitchen and a restaurant. There's a difference between a fab and a foundry. I could spin up a fab, no doubt, just like I can spin up a kitchen, but it won't be a good restaurant. You could spin up a fab, but it won't be a good foundry. The business is not easy, what TSMC does for a living is not easy, and it's not gonna get any easier.