GPU-96xCPU-1920GB-8xH100 by UpCloud

A high-density GPU instance featuring eight NVIDIA Hopper H100 GPUs and 1920 GB RAM for massive parallel computing workloads.

UpCloud GPU-96xCPU-1920GB-8xH100 is a high-density GPU-accelerated server designed for intensive parallel computing. Operating on the KVM hypervisor with an x86_64 CPU architecture, the instance features 96 shared vCPUs and 1920.0 GB of system memory, yielding 20.0 GB of RAM per core. The core of its performance lies in eight NVIDIA Hopper H100 GPUs, delivering a total of 640 GB of VRAM. No local storage is bundled with this instance, requiring external storage solutions for data persistence. One complimentary public IPv4 address is included. Given the shared CPU allocation, this server is optimized for tasks where the primary computational burden is offloaded to the GPUs. It is highly suited for large-scale machine learning, deep learning model training, and complex data science workloads.