NVIDIA's next-gen AD102 GPU: 76.3B transistors, up to 18432 CUDA cores

NVIDIA might have unveiled its next-gen Ada Lovelace GPU architecture during its GeForce Beyond broadcast, where we were introduced to a trio of new GeForce RTX 40 series graphics cards, but now we are getting more details on the new flagship AD102 GPU.

The next-gen NVIDIA AD102 GPU will be at the center of the GeForce RTX 4090 graphics card, with AD102 packing up to 76.3 billion transistors, with a die size of 608.4 mm². NVIDIA's new Ada Lovelace GPUs are being made on TSMC's new 4N process node, with 2.78x the transistor density of its current-gen GA102 "Ampere" GPU made on Samsung's custom 8nm process node.

3

The next-gen NVIDIA AD102 "Ada Lovelace" GPU, made on TSMC's new 4N process node

NVIDIA's new AD102 GPU has up to 18432 CUDA cores, up to 568 4th Gen Tensor Cores, and up to 142 3rd Gen RT cores while NVIDIA's new Ada Lovelace GPU architecture features Shader Execution Reordering (SER) which makes rasterization operations much, much faster: offering up to 25% more gaming performance.

• Read more: NVIDIA details its new GeForce RTX 4090 Founders Edition graphics card
• Read more: NVIDIA's new GeForce RTX 4090: 2x faster than RTX 3090 Ti for $1599
• Read more: NVIDIA's new GeForce RTX 4080 revealed: 16GB model starts at $1199
• Read more: NVIDIA DLSS 3 announced: 4x faster in Cyberpunk 2077 on Ada Lovelace

There are multiple AD102 GPUs with AD102-300 inside of the GeForce RTX 4090 packing 16834 CUDA cores, which means the full-fat AD102 GPU with up to 18432 CUDA cores means there are 2048 CUDA cores for NVIDIA to play with to introduce a higher-end GeForce RTX 4090 Ti or what I'd love to see: TITAN Ada with 48GB of GDDR6X memory.

3

The full specs on NVIDIA's new full-fat AD102 "Ada Lovelace" GPU