Greg Ernst, Intel | HPE Discover 2023
summary
Intel says integrated accelerators in Xeon deliver 10x improvement on PyTorch and TensorFlow versus prior gen, while GPU Max powers exascale clusters (six GPUs per node at Argonne). Video and edge workloads use the OpenVINO open‑source stack. Manufacturing targets include Five nodes in four years and a shift to move half global semiconductor capacity toward the western hemisphere, with 93% of products now made using 93% reusable energy and positive water reuse.
AI-written summary · about
