NVIDIA B200 Tensor Core GPU 192GB
NVIDIA B200 Tensor Core GPU 192GB is a high-performance AI accelerator built on the NVIDIA Blackwell architecture, designed for large-scale AI training and inference. Featuring 192GB HBM3e memory, up to 8 TB/s memory bandwidth, and advanced Tensor Cores with FP4 and FP8 support, it delivers exceptional performance for generative AI, large language models (LLMs), and HPC workloads in modern data centers.
Availability subject to compliance verification.
Specification | NVIDIA B200 Tensor Core GPU 192GB |
|---|---|
Architecture | Blackwell (GB200) |
Transistors | 208 billion |
Memory | 192GB HBM3e |
Memory Bandwidth | Up to 8 TB/s |
FP4 Tensor Core | Yes (Native) |
FP8 Tensor Core | Yes (2nd Generation) |
Transformer Engine | 2nd Generation |
NVLink | 5th Generation (Up to 1.8 TB/s) |
TDP | Up to 1000W |
Manufacturing Process | TSMC 4NP |