How NVIDIA DSX MaxLPS Maximizes AI Factory Throughput and Efficiency

За 30 секунд

Что произошло

NVIDIA DSX MaxLPS dynamically reallocates power across GPUs within a fixed budget, enabling 40% more GPUs and 49% higher throughput in the same power envelope.

Почему это важно

Data center operators managing AI inference workloads who want to increase GPU capacity without expanding power infrastructure or budgets.

На что обратить внимание

P99 latency increased 17%, median latency stayed stable. Tail latency degradation and telemetry reliability are critical factors operators must validate before production deployment.