Nvidia has acquired the American SchedMD, the team behind Slurm, a tool that acts as a dispatcher in large server farms: it queues tasks, allocates GPUs and processors to them, assigns priorities, and ensures that resources are not idle during model training and other heavy calculations.
Nvidia has acquired the American SchedMD, the team behind Slurm, a tool that acts as a dispatcher in large server farms: it queues tasks, allocates GPUs and processors to them, assigns priorities, and ensures that resources are not idle during model training and other heavy calculations.
According to Reuters, the amount of the deal was not disclosed. Nvidia said it would continue to develop and distribute Slurm as an open and vendor-neutral project, and access to the code would remain free. SchedMD, in turn, earned money mainly from engineering and technical support for companies that use Slurm in their own infrastructures.
Reuters lists cloud provider CoreWeave and the Barcelona Supercomputing Center as Slurm users. These customers run computing tasks that can eat up a significant portion of a data center’s capacity, and it’s critical for them to have predictable resource allocation.
The deal underscores Nvidia’s commitment not just to GPUs, but also to the software that drives how those GPUs work in production. Reuters recalls: CUDA has become the standard for most developers and is one of the key factors in Nvidia’s dominance. On the same day, the company introduced new open AI models, positioning them as faster and cheaper against the backdrop of competition, including from Chinese labs.
SchedMD was founded in 2010 by Slurm developers Morris «Moe» Jett and Danny Oble in Livermore, California. The company has approximately 40 employees.
Previously, dev.ua wrote about how Nvidia developed location verification technology that can indicate which country its chips are operating in. The company noted that this step will help prevent the smuggling of artificial intelligence chips to countries where their export is prohibited.
