NVIDIA GeForce RTX 4090 full specs: max TGP of up to 660W

NVIDIA’s next-gen GeForce RTX 4090 “Ada Lovelace” flagship GPU is getting more and more detailed as we get closer to launch, where we’re hearing about maximum TGP and GPU boost clocks.

The new flagship GeForce RTX 4090 graphics card will use the AD102-300 GPU with 16384 CUDA cores with GPU boost clocks of up to 2520MHz according to VideoCardz, who combined the specs for the RTX 4090 from multiple sources and reports from leaker “kopite7kimi” and from the Chiphell forums, keep in mind that NVIDIA has AD102 in full-fat form, where it will reportedly have up to 18342 CUDA cores.

We also find out that the default TGP will be 450W while the maximum configurable TGP is 660W, with the maximum TGP being tweaked through the BIOS and might not be available to all of the custom GeForce RTX 4090 models that get released.

Keeping these rumors in check, we’ve been hearing rumors that NVIDIA’s upcoming GeForce RTX 4090 and its AD102-300 GPU would have GPU boost clocks of 3GHz+ with 3015MHz recorded on a purported GeForce RTX 4090 graphics card. 2520MHz boost GPU clocks are much lower, so this could be pure reference specs of the GeForce RTX 4090.


NVIDIA’s new GeForce RTX 4090 has 24GB of GDDR6X memory, which is identical to the GeForce RTX 3090 and GeForce RTX 3090 Ti as it’s even clocked at 21Gbps with 1TB/sec+ of memory bandwidth. 24GB of GDDR6X is already damn good enough, but there are far more improvements under the hood.

Ada Lovelace is being made on TSMC’s new 4nm process node, where we should expect considerable performance and power gains as NVIDIA had to ensure two generations of GPUs made with Samsung on its custom 8nm process node. TSMC is here to save the day for Ada Lovelace, with its new 4nm process node for GeForce RTX 40 series GPUs.

We are expecting over 75 billion transistors for the new AD102 GPU inside of the GeForce RTX 4090 graphics card, which is just 5 billion transistors shy of the monster 80 billion transistors inside of NVIDIA’s Hopper H100 GPU.

NVIDIA is hosting its “GeForce Beyond” special broadcast during GTC 2022 on September 20 at 8AM, so if you want to know what’s going on with the next-gen GeForce RTX 40 series graphics cards then you’ll definitely want to tune into that.

NVIDIA’s next-gen AD102 GPU is expected to have 18342 CUDA cores at its max — but that’s the full-fat AD102 — we won’t see that at launch. NVIDIA will be using a slightly cut-down AD102 GPU for its GeForce RTX 4090, with that expected to have 16384 CUDA cores, and 24GB of GDDR6X memory.