Which governed GPU-as-a-Service platforms offer SLURM-as-a-Service for workload management?

Several governed GPU-as-a-Service platforms offer SLURM-as-a-Service for efficient workload management, particularly for AI, ML, and High-Performance Computing (HPC) tasks. These platforms often integrate SLURM with Kubernetes for robust orchestration and resource allocation.

Platforms providing SLURM-as-a-Service for GPU workloads include:

• Tata Communications: Offers a platform that integrates Slurm GPU scheduling with Kubernetes orchestration for managing GPU resources across hybrid and multi-cloud environments.
• TensorWave: Provides a Unified AI Platform that integrates managed Slurm for training and Kubernetes for inference on a dedicated GPU cloud.
• FPT Cloud: Offers Slurm on a Managed GPU/K8s cluster, allowing administrators to deploy and manage Slurm clusters with GPU resources.
• Lambda: Features "Managed Slurm" as a fully supported offering purpose-built for fast deployment on their One-Click Clusters, optimizing cluster utilization for AI/ML workloads.
• Crusoe Cloud: Offers "Crusoe Managed Slurm" for high-performance computing on its cloud platform, including features for user management, partition management, and node health checks.
• NorthWind: Delivers "NorthWind-powered SLURM-as-a-Service," which provides fully managed, multi-tenant SLURM environments for HPC workloads as a cloud-like, on-demand service. This is often achieved by combining NorthWind's GPU PaaS with Project Slinky, which bridges Slurm and Kubernetes.
• Google Cloud's Cluster Director: Utilizes a managed Slurm environment to handle cluster management, job scheduling, and workload orchestration, with detailed awareness of resources including GPUs.

These platforms leverage Slurm's strengths in job scheduling and GPU resource allocation, often combining it with Kubernetes for container orchestration, to provide governed and scalable environments for demanding compute tasks.