The NVIDIA GeForce RTX 4090 is the flagship Ada Lovelace architecture consumer GPU, launched in October 2022. It represents a generational leap in consumer graphics performance and has become popular not only for gaming but also for local AI inference workloads.
The RTX 4090 uses the full AD102 die manufactured on TSMC's custom 4N process with 76.3 billion transistors. It features 24GB of GDDR6X memory with 1 TB/s bandwidth. The chip includes 16,384 CUDA cores, 512 fourth-generation Tensor Cores, and 128 third-generation RT cores.
Fourth-generation Tensor Cores enable DLSS 3 with Frame Generation, a breakthrough feature that uses AI to generate entire frames between rendered frames, dramatically increasing frame rates in supported games. This technology fundamentally changed the performance equation for high-end gaming.
The RTX 4090 has become popular for local AI inference, capable of running large language models with billions of parameters. The 24GB memory capacity enables running models like Llama 2 13B locally. TDP is 450W, requiring robust power delivery. It defined the Ada Lovelace generation and remains the performance leader until Blackwell.