Tuesday, March 19th 2024
NVIDIA "Blackwell" GeForce RTX to Feature Same 5nm-based TSMC 4N Foundry Node as GB100 AI GPU
Following Monday's blockbuster announcements of the "Blackwell" architecture and NVIDIA's B100, B200, and GB200 AI GPUs, all eyes are now on its client graphics derivatives, or the GeForce RTX GPUs that implement "Blackwell" as a graphics architecture. Leading the effort will be the new GB202 ASIC, a successor to the AD102 powering the current RTX 4090. This will be NVIDIA's biggest GPU with raster graphics and ray tracing capabilities. The GB202 is rumored to be followed by the GB203 in the premium segment, the GB205 a notch lower, and the GB206 further down the stack. Kopite7kimi, a reliable source with NVIDIA leaks, says that the GB202 silicon will be built on the same TSMC 4N foundry node as the GB100.
TSMC 4N is a derivative of the company's mainline N4P node, the "N" in 4N stands for NVIDIA. This is a nodelet that TSMC designed with optimization for NVIDIA SoCs. TSMC still considers the 4N as a derivative of the 5 nm EUV node. There is very little public information on the power- and transistor density improvements of the TSMC 4N over TSMC N5. For reference, the N4P, which TSMC regards as a 5 nm derivative, offers a 6% transistor-density improvement, and a 22% power efficiency improvement. In related news, Kopite7kimi says that with "Blackwell," NVIDIA is focusing on enlarging the L1 caches of the streaming multiprocessors (SM), which suggests a design focus on increasing the performance at an SM-level.
Sources:
Kopite7kimi (Twitter), #2, VideoCardz
TSMC 4N is a derivative of the company's mainline N4P node, the "N" in 4N stands for NVIDIA. This is a nodelet that TSMC designed with optimization for NVIDIA SoCs. TSMC still considers the 4N as a derivative of the 5 nm EUV node. There is very little public information on the power- and transistor density improvements of the TSMC 4N over TSMC N5. For reference, the N4P, which TSMC regards as a 5 nm derivative, offers a 6% transistor-density improvement, and a 22% power efficiency improvement. In related news, Kopite7kimi says that with "Blackwell," NVIDIA is focusing on enlarging the L1 caches of the streaming multiprocessors (SM), which suggests a design focus on increasing the performance at an SM-level.
60 Comments on NVIDIA "Blackwell" GeForce RTX to Feature Same 5nm-based TSMC 4N Foundry Node as GB100 AI GPU
Was Maxwell a failed product? Was Fermi 2.0? Was second gen Kepler? Was Turing, for that matter, since it was just an optimization of the 16nm that Pascal used? If you expect every gen to be a massive uplift like Pascal was… well, that’s on you. Hell, look at Ampere. Those cards were good despite the new process, not because of it. And yeah, blame lies with Samsung there. But NV cannot control TSMC either. From what I understand, 3nm is just straight up not ready for prime time yet, there are yield issues. And in the current market where demand is high the last thing NV needs is yield issues.
Seriously, people get hung up on fabrication processes too much.
Maybe they'll take advantage of new features on the software side like "graphs" to show bigger numbers, or maybe they'll just cross all the efficiency lines to deliver more performance.
How the actual hell have they failed? So 4 years on the same process was all kosher. Then another 4 years on the same process - still fine (as I said, 12nm TSMC IS just an optimization of 16nm). But now suddenly a second gen in only two years using the same process is a failure? I fail to see your logic. And how has NV failed here, even? Again, they work with what TSMC has. If 3nm isn’t ready then it isn’t ready. What, do you expect Raptor Jesus to come on down from heavens and make it work for them? Or should they delay the chips indefinitely until it DOES work and then pour more resources into shrinking the arch onto new node? Do you understand how these things even work?
NVIDIA will probably still be secure even with the RTX 4090 vs RDNA4, let alone anything RTX 50xx.
Without a node upgrade I'd expect ~25-40% improvements just from architecture/more cores, maybe with some more unique hardware accelerators like the optical flow unit in Ada, for example. Perhaps in some scenarios a ~50% performance improvement (ray tracing?) like the leaks suggest.
Apple will as usual be buying all the 3nm leading edge stuff, so why fix something that isn't broken?
Copium, I know... but there's a chance!
$1000 5080 would be nice, and possible, unlike the $600 5080 some people seem to expect/want.
On that point, I have no doubt NVIDIA will be able to extract enough performance from the node.
"Here gamers, take the old A.I chips we couldn't sell, lol, now give me $5000 / GPU." :roll:
2020, RTX 3080 - $700
2022, RTX 4080 - $1200
2024, RTX 5080 - $2060
Second, it's not happening. Not if they're planning to increase the front end's L1 cache.
The increase on the L1 cache shows their Shaders/or other parts are just sitting idle & still not being used mostly.
The best Nvidia can do right now is 20% increase in performance compared to the 4090 at the same power.]
Raytracing side hasn't increase more than 6% pre-clock compared to each generation. The only time Raytracing has had a major increase on Nvidia is when rasterization was massively increase above it.