In reality, it’s even more, because the RTX 4090 doesn’t even utilize the full die — the GPU serves up 16,384 CUDA cores, but the chip has up to 18,432.

Source: https://www.digitaltrends.com/computing/nvidia-rtx-5090-specs-early-leak/

Lavender said the tool is typically able to migrate 90% of CUDA source code automatically to the C++ source code, leaving very little for programmers to tune manually.

Source: https://www.networkworld.com/article/3692348/intel-announces-144-core-xeon-processor.html

Note, however, that the mobile GeForce RTX 4090 is based on the AD103 GPU, similar to a desktop-class RTX 4080, and packs 9,728 CUDA cores.

Source: https://hothardware.com/reviews/razer-blade-16-with-rtx-4090-and-13th-gen-review

The RTX 3500 is stacked right in the middle of the entire lineup, featuring 5120 CUDA cores, 40 RT cores, and 160 Tensor cores, operating with up to a 140W TGP and 12GB of memory featuring 432GB/s of bandwidth.

Source: https://hothardware.com/news/nvidia-gtc-2023-unveils-ada-rtx-4000-sff-pro-laptop-gpus

These have configurable TDPs of 10 to 25 Watts, a full 102GB/s of memory bandwidth, and a Volta-based GPU with 1,024 CUDA cores and 32 Tensor cores.

Source: https://hothardware.com/reviews/nvidia-jetson-orin-nano-hands-on