NVIDIA's GeForce RTX 40 Super GPU Series Teaser Is Out Of This World
There's been a ton of unofficial buzz surrounding NVIDIA's purported GeForce RTX 40 series refresh with upcoming 'Super' models. While there's been no official word from NVIDIA just yet—that will probably come on or around the Consumer Electronics Show (CES) next week—the GPU designer appears to be teasing the expected launch with updated social media cover photos.
In the past 24 hours, NVIDIA changed its Facebook and Twitter cover photos for its 'NVIDIA GeForce' accounts, which both now show a shadowy image of a graphics card hovering in space. The backdrop, as can be seen above, is of planet Earth and a spattering of stars. We'd have gone with a supernova as the backdrop, but perhaps NVIDIA wanted to be more subtle.
Of course, NVIDIA CEO Jensen Huang has a history of throwing last-minute curve balls, though he usually reserves those pitches for major new architectures and accompanying GeForce products. What's presumably on tap in the more immediate future is a trio of upgraded Ada Lovelace cards with the Super branding attached. Specifically, the rumor mill has been buzzing about three models: GeForce RTX 4080 Super, GeForce RTX 4070 Ti Super, and the GeForce RTX 4070 Super.
Here's a high-level look at the leaked specs and how they compare to the non-Super variants...
- GeForce RTX 4080 Super: 10,240 CUDA cores, 2,550MHz boost clock, 16GB GDDR6X, 256-bit memory bus, 736GB/s memory bandwidth
- GeForce RTX 4080: 9,728 CUDA cores, 2,505MHz boost clock, 16GB GDDR6X, 256-bit memory bus, 716.8GB/s memory bandwidth
- GeForce RTX 4070 Ti Super: 8,448 CUDA cores, 2,610MHz boost clock, 16GB GDDR6X, 256-bit memory bus, 672GB/s memory bandwidth
- GeForce RTX 4070 Ti: 7,680 CUDA cores, 2,610MHz boost block, 12GB GDDR6X, 192-bit memory bus, 504.2GB/s memory bandwidth
- GeForce RTX 4070 Super: 7,168 CUDA cores, 2,475MHz boost clock, 12GB GDDR6X, 192-bit memory bus, 504GB/s memory bandwidth
- GeForce RTX 4070: 5,888 CUDA cores, 2,475MHz boost clock, 12GB GDDR6X, 192-bit memory bus, 504.2GB/s memory bandwidth
The Super specs are not official but if the leaks hold true, each model should offer a decent bump in performance over their non-Super counterparts. Each one is tipped for an increase in CUDA cores—21.7% more on the GeForce RTX 4070 Super, 10% more on the GeForce RTX 4070 Ti (which also gets a bump in VRAM capacity and speed), and 5.3% on the GeForce RTX 4080 Super.
It will also be interesting to see how NVIDIA prices its upcoming cards. The best-case realistic scenario is that NVIDIA retains the same MSRPs for the Super upgrades, while cutting the price of the non-Super models. We'll have to wait and see if that happens, and probably not very long based on its updated cover photos.
