pair of GeForce RTX 4080 models that are going to be vying for your wallet.
We have the release dates, pricing, a hint at the performance, and the specifications for the pair of RTX 4080 GPUs. Like the GeForce RTX 4090, these are built on Nvidia's latest Ada Lovelace architecture. Thanks to the use of TSMC's N4 production process—the same one that powers the GPUs in HPC data centers—these cards are looking at a superb boost to efficiency and performance over the previous generation of cards.
Despite these cards sharing the RTX 4080 name, they are very different when it comes to their internals, and not just down to the amount of GDDR6X VRAM they sport.
The GeForce RTX 4090 will be the first out of the door on October 12, but this will be followed up by the pair of GeForce RTX 4080s in November. You'll find the full specs and details of all these GPUs below.
The GeForce RTX 4080 (16GB) and GeForce RTX 4080 (12GB) cards are “Coming November,” with no exact date shown as yet. Once the RTX 4090 hits the shelves on October 12, we expect a firmer date for the two RTX 4080s.
You can sign up to be notified when the cards are available over on Nvidia's site. Resellers are expected to offer a similar service—not quite pre-orders, allowing buyers to sign up for notifications as to when the cards arrive. Given the lack of a hard release date, this does make sense here.
There's nothing official from Nvidia on when the review embargo lifts for these cards, which isn't too surprising given the lack of a release date.
When it comes to the thorny subject of price, the two RTX 4080s make for difficult reading. While the top-end, the GeForce RTX 4090, can naturally enjoy an elevated price point due to its flagship positioning, that's not usually true when it comes to the 80-level cards.
The GeForce RTX 4080 (16GB) has a starting price of $1,199. For comparison, the GeForce RTX 3080 (10GB) launched at $699, making this $500 more expensive offering. Like the RTX 4090, this will be available as a Founders Edition card.
The GeForce RTX 4080 (12GB) has a starting price of $899. There's no Founders Edition version of this card though, so it'll be down to the graphics card manufacturers to try and hit that price point. With elaborate cooling options and no doubt improved overclocking chops, we'd expect plenty of cards to cost more than this.
Nvidia has released some internal RTX 4080 benchmark figures to show off its latest architecture, and it certainly makes for some interesting reading. While all of the new Ada Lovelace GPUs enjoy significant increases in what Nvidia calls “Next Generation” games, things aren't looking so rosy when it comes to today's games.
The likes of Resident Evil Village, Assassin's Creed Valhalla, and The Division 2 all appear to show performance drops for the RTX 4080 (12GB) edition and only small increases for the RTX 4080 (16GB) compared to the RTX 3090 Ti.
This isn't the absolute home run you might expect for the money, though things are a bit healthier when it comes to Microsoft Flight Simulator and Warhammer 40,000: Darktide, which enjoy an almost doubling of the frame rate. It's going to be interesting to see how these cards perform in a range of titles when we do get them on the test benches—because while the RTX 4090 clearly has some serious performance chops, it isn't as clear cut lower down the stack.
The two RTX 4080s use Nvidia's Ada Lovelace architecture, boast support for the new 3rd generation Ray tracing cores, 4th generation Tensor cores, and DLSS 3. Nvidia has stuck with the PCIe 4.0 interface for this generation and the cards support Resizable BAR, Microsoft DirectX 12 Ultimate, and Vulkan RT API.
Nvidia will be producing Founder Editions of the GeForce RTX 4080 (16GB) cards, but not of the GeForce RTX 4080 (12GB), which means the specs for the $899 cards will be entirely down to the manufacturers when it comes to fan configurations and the overall size of the cards.
As I said at the beginning, there are a lot of differences between the RTX 4080 (12GB) and the RTX 4080 (16GB), far more than just the amount of memory available. The most obvious of these is the number of CUDA Cores—9,728 vs. 7,680 is not a small difference.
This is a very different situation from the previous generation's RTX 3080 (10GB) model and the RTX 3080 (12GB) GPU that was released at the start of the year, which had 8,704 cores and 8,960 respectively.
You could absolutely make the case that the GeForce RTX 4080 (12GB) should have been called the RTX 4070.
|RTX 3080 (10GB)||RTX 3080 (12GB)||RTX 4080 (12GB)||RTX 4080 (16GB)|
|Memory Type||10GB GDDR6X||12GB GDDR6X||12GB GDDR6X||16GB GDDR6X|
|Graphics Card Power (W)||320W||350W||285W||320W|
|Required System power (W)||750W||750W||700W||750W|