NVIDIA Quadro RTX 4000 8GB vs NVIDIA GeForce RTX 2080 Ti 11GB

Both GPUs launched in 2018: the Quadro RTX 4000 on Turing and the GeForce RTX 2080 Ti on Turing.

Dense throughput for the Quadro RTX 4000 against the GeForce RTX 2080 Ti: FP64 0.2 vs 0.4 TFLOPS, FP32 7.1 vs 14 TFLOPS, FP16 57 vs 57 TFLOPS. Sparse figures, where the vendor publishes them, appear under each dense number in the performance table.

Memory is 8 GB against 11 GB, fed at 416 GB/s versus 616 GB/s. More memory per GPU means larger models fit before you have to shard across cards, which often matters more than raw TFLOPS for inference.

Power budgets are 160 W for the Quadro RTX 4000 and 260 W for the GeForce RTX 2080 Ti. At FP32 that works out to 0.04 against 0.05 TFLOPS per watt.

For LLM training and inference, weight the FP8 and FP16 rows and memory capacity most heavily. For scientific and HPC workloads, the FP64 row is the one to read.