NVIDIA Tesla P100 16 GB HBM2 graphics card

Data-center accelerator for HPC and deep learning

The NVIDIA Tesla P100 16 GB HBM2 is a plug-in card built for data-centre workloads such as high-performance computing and deep learning. It targets researchers and engineers who need maximum throughput for large-scale simulations and model training. The GPU delivers the compute density required to replace many commodity nodes with fewer accelerators.

256-Bit HBM2 memory interface for dense builds

The 16 GB HBM2 memory on a 256-Bit interface provides high bandwidth in a compact form factor that fits standard server slots. This density frees rack space and reduces interconnect complexity compared with clusters of smaller cards. The design lets administrators scale capacity without expanding the physical footprint of the compute cluster.

Sustained throughput for long-running workloads

The architecture is engineered for continuous operation across extended HPC and deep-learning runs rather than short bursts. Over 400 HPC applications are accelerated, including 9 out of 10 of the most widely used codes. This breadth of support keeps utilisation high during multi-day simulations and large-batch training jobs.