The NVIDIA H100 80GB PCIe GPU is a powerful solution for AI and high-performance computing (HPC) workloads. It features 14592 CUDA cores and 456 fourth-generation Tensor cores, delivering up to 51.22 TFLOPS in FP32 performance and 204.9 TFLOPS in FP16 performance. The GPU is equipped with 80 GB of HBM2e memory, offering a peak memory bandwidth of 2.04 TB/s. It supports Multi-Instance GPU (MIG) technology, allowing the GPU to be partitioned into up to seven smaller, fully isolated instances, each with its own dedicated resources such as memory and compute cores.
