High Performance Computing (HPC)

High Performance Computing (HPC)#

The Concept

  • Cluster: Aggregation of individual computers

  • Interconnected: High-speed (internal) networking

  • Access: Streamlined interface

The Goal

  • Scalability: Tackle larger workloads

  • Performance: Solve problems faster

  • Efficiency: Optimize hardware utilization

  • Equitable Access: Ensure balanced resource sharing

Multi-Tenant

  • You are not alone!

  • Resources are finite.

  • “Idle” jobs block others.

  • Unused or paused VM’s deprive resources.

Fair Share (Cluster)

  • Priority is dynamic.

  • Heavy users get lower priority over time.

  • Accurate time estimates = happier scheduler.

Walltime

The Walltime is the maximum real-world time your job is allowed to run. If your job hits this limit, the scheduler kills it immediately. Always overestimate slightly (e.g., +20%).