The GeForce RTX 40 series was finally revealed, but in keeping with Nvidia’s true style, CEO Jensen Huang did not delve deep into the raw speeds and channels during the RTX 4090 and 4080 presentation during his GTC speech. After all, there is only so much time for these events, and mass audiences do not like sifting through verbal descriptions of technical footnotes. Fortunately, Nvidia’s website has some key technical details that are missing from the presentation – details that confirm some very interesting tidbits about these new GPU-powered “Ada Lovelace” graphics cards.
Here are six details about GeForce RTX 40-series cards you need to know no hear during Jensen’s disclosure. For full details of Nvidia’s new graphics cards, check out our coverage of the GeForce RTX 4090 and 4080 announcement.
One name, two very different RTX 4080s
In a deeply anti-consumer move, Nvidia launches two GeForce RTX 4080 cards, seemingly differing in memory capacity. Nvidia simply named them “16 GB RTX 4080” and “12 GB RTX 4080”. But while Huang didn’t mention it, these two “4080s” will also offer incredibly different base performance.
As you can see in the chart above, the 12GB GeForce RTX 4080 not only has less memory, but also a much narrower 192-bit bus and an amazing 21 percent less CUDA cores. This means it will be much slower than the 16GB model, more often than not. Naming these two very different GPUs with the same name is sure to cause confusion among buyers of graphics cards.
The 12 GB RTX 4080 will not have the Founders Edition model
Speaking of which, a scroll of the specs sheet for the RTX 4080 revealed this interesting additional curiosity. While the 16GB GeForce RTX 4080 includes specific dimensions for length, width, and slots for Nvidia’s Founders Edition, the 12GB GeForce RTX 4080 simply says “varies by manufacturer.” It doesn’t look like the 12GB model will get its own custom Nvidia Founders Edition. Update: Brian Burke from Nvidia confirmed to PCWorld that there will be no Founders Edition 12GB RTX 4080 version.
GeForce RTX 40 series supports AV1 encoding
AV1 encoding is the holy grail for multimedia creators, offering much better visuals with much lower bandwidth needs. Intel beat AMD and Nvidia to a blow by incorporating killer AV1 encoding into their Arc graphics cards, but footnotes in the Nvidia RTX 4090 and 4080 spec sheets reveal that Team Green will support this generation AV1 encoding as well (AV1 decoding was already supported).
“GeForce RTX 4090 and GeForce RTX 4080 graphics cards feature two of our new eighth generation (NVENC) NVIDIA encoders, now with AV1 encoding support, bringing tons of new possibilities for streamers, video editors, and video calling.” – RTX 40 – announcements post series crowed.
You read it right – not one but two NVENC encoders with support for AV1 encoding on RTX 40 series graphics cards. Pay attention to streamers.
DLSS 3 will not appear on older GPUs
It’s obvious that the last generation GPUs will not include next-generation hardware encoders. But the very popular Nvidia DLSS 3 software feature will also not appear on the RTX 20 and 30 series cards, even though DLSS 2.0 and 2.1 have been ported to both. Nvidia’s GPU generation comparison page lists the DLSS 3 only on the RTX 40 series, while the older generation RTX offerings remain on the list with DLSS 2.0.
This is because DLSS 3 software requires hardware that older GPUs just don’t have. Nvidia spokesman Brian Burke explained the situation of PCWorld via email:
“DLSS 3 consists of 3 technologies – DLSS Frame Generation, DLSS Super Resolution (also known as DLSS 2) and NVIDIA Reflex.
DLSS frame generation uses the RTX 40 series high-speed optical flow accelerator to calculate the traffic flow used in the AI network, and then executes the network on fourth generation tensor cores. Support for previous GPU architectures would require further innovation and optimization of the optical flow algorithm and the AI model.
DLSS Super Resolution and NVIDIA Reflex will of course continue to be supported on previous generation hardware, so current GeForce gamers and developers will benefit from DLSS 3 integrating games. We will continue to research and train AI for DLSS Super Resolution and provide model updates for all RTXs. customers as we’ve been doing since the first DLSS release. For the vast majority of game integrations, both DLSS Super Resolution and Frame Generations will appear as separate UI settings, and the player can choose one, the other, or both. “
You’re gonna need a bigger boat
The GeForce RTX 4090 doesn’t drain 600 watts plus, as some early rumors suggested – at least not in Nvidia’s Founders Edition iteration – but if have a power rating of 450 W, a power supply of at least 850 W is recommended. This fits the flagship RTX 3090 Ti from the last generation. As we said then: you’ll need a bigger boat. And by boat, we mean power. However, this is not all expensive news. While the RTX 40 series supports the 12VHPWR-pin connectors found in the new ATX 3.0 power supplies, you can also use three or four standard 8-pin power connectors if you already have a powerful power supply.
NVLink is dead for GeForce
Nvidia SLI technology for multi-GPU configurations has been down for a while, but the “NVLink” technology that replaced it remained in the last-gen RTX 3090 GPUs. Not any more. The spec pages for the GeForce RTX 4090 and 4080s clearly say NVLink is no he supported this generation. Pour one out.
So much for these buried treasures. We’ll be sure to find out more about the GeForce RTX 40 series in the days and weeks to come, but once again, to learn more about Nvidia’s new graphics cards, check out our coverage of the GeForce RTX 4090 and 4080 announcement.
Editor’s Note: This article has been updated with comments by Brian Burke from Nvidia.