Intel Deep Learning (DL) Boost with BFloat16 (BF16) demonstrates benefits across deep learning training workloads with the same accuracy as 32-bit floating-point (single-precision) (FP32). Recently Amazon introduced EC2 M6i instances powered by the latest-generation Intel Xeon Scalable Processors. Intel and Alluxio collaborate to measure a 20-25% price/performance improvement over the prior generation for machine learning models with PyTorch on AWS. This collaboration demonstrates data preprocessing and training at lower cost on CPUs using Alluxio as the data access layer to cloud storage.
Solution Brief