NVIDIA GB200 Grace Blackwell Superchip 372GB vs NVIDIA B200 SXM 180GB

Both GPUs launched in 2024: the GB200 on Blackwell and the B200 on Blackwell.

Dense throughput for the GB200 against the B200: FP64 80 vs 37 TFLOPS, FP32 160 vs 75 TFLOPS, FP16 5,000 vs 2,250 TFLOPS, FP8 10,000 vs 4,500 TFLOPS. Sparse figures, where the vendor publishes them, appear under each dense number in the performance table.

Memory is 372 GB against 180 GB, fed at 16.0 TB/s versus 7.7 TB/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.

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.