High capacity for data centers
Intel has introduced its new D-series GPU aimed at AI inference. Its standout feature is up to 480 GB of LPDDR5X memory.
Memory capacity is critical for running large models on a single card. LPDDR5X offers high bandwidth, though its cost and power profile differs from HBM.
Who is it for?
Intel is targeting data centers. By emphasizing high memory capacity, the company aims to make it easier to host larger models for inference workloads.
The source text does not include pricing, launch date or performance figures. Those details are expected to become clearer in the coming period.
