System Overview#
Launched in October 2026, Calder provides dedicated CPU and GPU resources for a wide range of scientific computing workloads. The CPU partition consists of 96 compute nodes, providing a total of 12,288 CPU cores and approximately 73.7 TB of memory. The GPU partition consists of 7 compute nodes, providing a total of 56 NVIDIA H200 NVL GPUs and approximately 10.5 TB of memory. All systems are connected via a high-performance NDR InfiniBand 400 Gb/s network. Further details are provided in the sections below.
The Aire HPC system has been designed with a scalable architecture capable of supporting a wide variety of research applications. The system consists of 52 standard compute nodes and 2 high-memory compute nodes, each equipped with 168 cores, providing a total of 9,072 cores. It also includes 28 GPU nodes, each containing three NVIDIA L40S GPUs, offering a total of 84 GPUs. Aire provides a generous amount of local storage to support the demands of large-scale HPC workloads.
Note
Before diving into the Aire and Calder HPC systems, it’s important to have a basic understanding of their architecture and capabilities. This will help you make the most of the resources available and ensure efficient use of both systems.
Explore the sections below to learn more about our HPC privision: