NVIDIA scooped up SchedMD, the crew behind Slurm workload management software that handles job scheduling for more than half the systems on the TOP500 supercomputer rankings. The chipmaker promised to keep the project open source and vendor-neutral while throwing resources at development to help AI labs and research clusters run more efficiently.
Slurm gets used by foundation model developers to wrangle training runs and inference jobs across massive GPU farms. SchedMD CEO Danny Auble said teaming up with NVIDIA validates how critical the scheduler has become for cutting-edge compute infrastructure. The companies have already worked together for over a decade.
NVIDIA plans to give SchedMD faster access to new hardware while maintaining support for mixed vendor setups, letting customers optimize workloads across whatever gear they run.
Slurm gets used by foundation model developers to wrangle training runs and inference jobs across massive GPU farms. SchedMD CEO Danny Auble said teaming up with NVIDIA validates how critical the scheduler has become for cutting-edge compute infrastructure. The companies have already worked together for over a decade.
NVIDIA plans to give SchedMD faster access to new hardware while maintaining support for mixed vendor setups, letting customers optimize workloads across whatever gear they run.
