The NVIDIA Tesla V100 PCIe Tensor Core GPU is a mature Volta-generation data center compute accelerator, equipped with 32GB ECC HBM2 memory and 640 Tensor Cores. It supports FP16, FP32 and FP64 computing, delivering reliable performance for model inference, computational fluid dynamics, weather simulation and data analytics.
This PCIe passive-cooled dual-slot variant is easy to install in standard off-the-shelf rack servers, without requiring proprietary SXM HGX chassis. It supports MIG to split the GPU into multiple isolated compute instances for multi-user resource sharing.
It has no native display outputs, designed purely for compute workloads. It is the preferred budget GPU for universities and small research groups with limited capital.