Marketplace · HPC & batch

HPC & Batch Compute on VirtEngine

Scheduler-backed capacity, bought from the catalogue. The listing describes the queue, the nodes, the GPUs and the job limits; the provider's scheduler runs the work.

Read the HPC guide For HPC facilities
Protocol in development Example listings are illustrative
HPC offering Queue / partition Lease Provider daemon SLURM / scheduler Job
The scheduler runs the job. The chain records the lease and settles signed usage.

Listing

What an HPC listing exposes

HPC offerings are catalogue products, not raw scheduler access. The provider publishes the constraints; the listing makes them visible before you order.

FieldWhat it fixes
Queue / partitionwhich scheduler lane the job enters
Node typeCPU class, GPU class, interconnect
CPU / GPU countsper-job or per-allocation limits
Wall timemaximum job duration before the queue ends it
Region / facilitywhere the cluster sits
Storagescratch, project or parallel filesystem

Catalogue

Example HPC listings

Three illustrative products — example providers, example figures.

HPC

Illustrative example

GPU Batch Queue

Example Provider

Example facility

  • 4 GPUs
  • 32 CPU cores
  • 6-hour maximum job

Billing components

  • GPU-hour
  • core-hour

Allocation-based

HPC

Illustrative example

CPU Cluster Allocation

Example Provider

Example facility

  • 256 core-hours
  • Shared queue
  • Priority: standard

Billing components

  • core-hour

Allocation-based

HPC

Illustrative example

MPI Job Template

Example Provider

Example facility

  • Interconnect-aware
  • Template defined
  • Node count configurable

Billing components

  • node-hour

Per-job pricing

More examples
  • Interactive Session — on-demand Jupyter node

Workloads

Typical HPC workloads

Difference

How it differs from generic VM IaaS

VM IaaS

You operate a machine: an OS, a network identity and everything above it, for as long as you keep it.

HPC listing

You submit work into a scheduler: queues, wall time and node classes replace machine ownership, and metering follows the job or allocation.

Full architecture: the HPC on VirtEngine guide covers schedulers, workload patterns and cluster operations; HPC cluster operators covers the provider side; the x/hpc module documents the job marketplace surface.

Continue

Related pages