NVIDIA's new GeForce RTX 4080 revealed: 16GB model starts at $1199

NVIDIA has finally announced its next-gen Ada Lovelace GPU architecture and with it three new GeForce RTX 40 series graphics cards: the flagship GeForce RTX 4090, and two new GeForce RTX 4080 graphics cards.

The new GeForce RTX 4080 is exactly as the rumors suggested: split into the RTX 4080 16GB and RTX 4080 12GB. NVIDIA's new GeForce RTX 4080 16GB features the AD103-300 GPU made on TSMC's new 4N process node, with 9728 CUDA cores, and a GPU boost clock of up to 2510MHz.

6

NVIDIA's new RTX 4080 has some freaking incredible gains over the RTX 3080 Ti

There's 16GB of GDDR6X memory on a 256-bit memory bus clocked at 22.5Gbps, offering up to 720GB/sec of memory bandwidth. NVIDIA's new GeForce RTX 4080 16GB graphics card has a default TGP of 320W.

Under that, we've got the GeForce RTX 4080 12GB that uses the AD104-400 GPU that is still made on the TSMC 4N process node but gets CUDA cores ripped away: just 7680 CUDA cores on the RTX 4080 12GB, while GPU boost clocks jump up to 2610MHz.

• Read more: NVIDIA's new GeForce RTX 4090: 2x faster than RTX 3090 Ti for $1599
• Read more: NVIDIA DLSS 3 announced: 4x faster in Cyberpunk 2077 on Ada Lovelace

NVIDIA is using 12GB of GDDR6X memory clocked at 21Gbps on a 192-bit memory bus that's offering up to 504GB/sec of memory bandwidth, with a lower 285W TGP.

6

NVIDIA's new GeForce RTX 4080, in all its next-gen Ada Lovelace GPU glory

NVIDIA will be launching the new GeForce RTX 4080 16GB and RTX 4080 12GB graphics cards in November starting at $899 for the RTX 4080 12GB and then starting at $1199 for the RTX 4080 16GB.

• Read more: NVIDIA's new GeForce RTX 4090: 2x faster than RTX 3090 Ti for $1599

NVIDIA's GeForce RTX 4080 is also built with creators in mind, with its additional memory -- 16GB up from the 10GB of G6X that launched on the RTX 3080 -- which will help out when rendering larger scenes, editing 12K RAW video, and working with multiple apps at the same time.

All of the new Ada Lovelace GPUs are compatible with NVIDIA DLSS 3, which is the very latest AI-powered upscaling technology ready for next-gen GeForce RTX 4090, RTX 4080 16GB, and RTX 4080 12GB graphics cards.

NVIDIA says its new DLSS 3 technology will super-power games like Cyberpunk 2077 with its new Ray Tracing mode "Overdrive Mode" with DLSS 3 offering 90FPS+ average with DLSS 3 + RT On, compared to just 20FPS or so average with DLSS disabled but RT still enabled. That's a 4x performance improvement, something that I think every gamer on the planet wants to see up and running.

Now I want to see what DLSS 3 can do for 8K gaming, which has been something that I've been waiting for... for a very, very long time. It seems DLSS 3 is ready for 8K 60FPS and maybe even 8K 120FPS gaming... oh thank you, Ada.

NVIDIA GeForce RTX 4090 24GB

• CUDA cores: 16384
• GPU boost clock: 2.52GHz
• GPU base clock: 2.23GHz
• Memory: 24GB GDDR6X
• Memory interface: 384-bit
• Graphics card power: 450W
• Power connectors: 3 x 8-pin (or PCIe Gen5 cable)

6

NVIDIA GeForce RTX 4080 16GB

• CUDA cores: 9728
• GPU boost clock: 2.51GHz
• GPU base clock: 2.21GHz
• Memory: 16GB GDDR6X
• Memory interface: 256-bit
• Graphics card power: 320W
• Power connectors: 3 x 8-pin (or PCIe Gen5 cable)

NVIDIA GeForce RTX 4080 12GB

• CUDA cores: 7680
• GPU boost clock: 2.61GHz
• GPU base clock: 2.31GHz
• Memory: 12GB GDDR6X
• Memory interface: 192-bit
• Graphics card power: 285W
• Power connectors: 2 x 8-pin (or PCIe Gen5 cable)

Tune in for custom AIB models of the GeForce RTX 4080 graphics cards, which will be coming from ASUS, COLORFUL, Inno3D, GAINWARD, GIGABYTE, MSI, Palit, ZOTAC, and others. There's no EVGA obviously, given what's happened in the last few days. It's a pity, given how much of a beast Ada Lovelace is shaping up to be.