Filtering data center power load transients caused by artificial intelligence (ai) workloads

A computer-implemented method, comprising:

• receiving, a workload orchestrator, a first signal indicating that a first plurality of compute nodes is starting a compute phase during which artificial intelligence (“AI”) workloads are executed by AI accelerators on the first plurality of compute nodes;
• in response to receiving the first signal, causing, by the workload orchestrator, a second plurality of compute nodes to stop execution of general workloads;
• receiving, by the workload orchestrator, a second signal indicating that the first plurality of compute nodes has completed the compute phase and is starting a communication phase during which AI data is exchanged among the AI accelerators on the first plurality of compute nodes;
• and
• in response to receiving the second signal, causing, by the workload orchestrator, the second plurality of compute nodes to continue execution of the general workloads.

A system, comprising:

• a workload orchestrator, comprising:
• an artificial intelligence (“AI”) workload scheduler that schedules AI workloads for AI accelerators on each of a first plurality of compute nodes to execute;
• a non-AI workload scheduler that schedules general workloads for graphics processing units (“GPUs”) on each of a second plurality of compute nodes to execute, wherein the first plurality of compute nodes is disposed on and electrically powered by a first plurality of equipment racks, wherein the second plurality of compute nodes is disposed on and electrically powered by a second plurality of equipment racks;
• and
• a rack/row controller system that allocates racks and rows for AI workloads and for the general workloads;
• wherein workload orchestrator that executes computer executable instructions that cause the workload orchestrator to perform operations comprising:
• receiving, by the rack/row controller system and from a power meter, a first signal indicating that a current power draw by at least the first plurality of compute nodes and the second plurality of compute nodes exceeds a power threshold value, indicative of the first plurality of compute nodes starting a compute phase during which the AI workloads are executed by the AI accelerators on the first plurality of compute nodes;
• in response to receiving the first signal, causing the second plurality of compute nodes to stop execution of the general workloads, by the rack/row controller system sending instructions to a power capper to instruct one or more power distribution units (“PDUs”) to throttle power feeding the second plurality of equipment racks to a first operational power level;
• receiving, by the rack/row controller system and from the power meter, a second signal indicating that a current power draw by at least the first plurality of compute nodes and the second plurality of compute nodes falls below the power threshold value, indicative of the first plurality of compute nodes having completed the compute phase and is starting a communication phase during which AI data is exchanged among the AI accelerators on the first plurality of compute nodes;
• and
• in response to receiving the second signal, causing the second plurality of compute nodes to continue execution of the general workloads, by the rack/row controller system sending instructions to the power capper to instruct the one or more PDUs to disable power throttling to the second plurality of equipment racks to set the power feeding the second plurality of equipment racks at a second operational power level.