반도체·하드웨어 · 리포트
SageMaker Inference now adds GPU partitioning with NVIDIA MIG to its inference service.
Finer-grained GPU slicing lets inference workloads share accelerators more efficiently, which can raise utilization and lower per-request cost on cloud GPU capacity.
업계 전문지Slicast · 2026년 11월 7일 10:19 UTC · 미국 · 출처: The Chenab Times
중요도 50