
AI
Microsoft targets idle GPUs with dynamic composable memory for AI data centers
Microsoft is scouting composable memory technologies to address the scarcity of memory resources in AI data centers and reduce idle GPU capacity.
Everything Vector Tribune has reported about AI inference: 2 stories, newest first. Part of our AI coverage.

Microsoft is scouting composable memory technologies to address the scarcity of memory resources in AI data centers and reduce idle GPU capacity.

The new custom accelerator uses a chiplet design and dataflow architecture to improve energy efficiency for large-scale AI workloads.