Nutanix is acquiring France-based AI computing orchestration technology company Ryax Technologies, Techzine reported on Sept. 23 local time.

Nutanix plans to integrate Ryax orchestration and scheduling technology into its next Nutanix Kubernetes Platform (NKP) and Nutanix Enterprise AI (NAI), the report said.

The aim is to help companies use GPUs more efficiently and automatically place AI workloads on the cheapest or most energy-efficient hardware.

Ryax functions to be included in NKP focus on analysing usage data. The system learns past resource-usage patterns and adjusts GPU allocations each time a job runs.

If a job fails due to insufficient memory during execution, the system allocates more GPU memory and can automatically restart the job. It also provides split GPU functions that let multiple containers share a single GPU, and serverless GPU functions that reserve resources only when actual computation is performed.

Nutanix said that in Ryax tests, applying per-run resource adjustments cut node usage time by 62 percent for running 30 deep-learning jobs and sped up completion by 5.7 percent.

Nutanix plans to deploy a global meta-scheduling layer in NAI to coordinate overall workloads. The layer allocates training, batch and inference tasks to meet performance, cost and energy goals, then sends them to Nvidia or AMD equipment, public cloud or Slurm-based HPC clusters. It scores each cluster using a hardware power model to place tasks where expected energy use is lowest, Techzine said.