Amazon FSx for Lustre now supports Elastic Fabric Adapter (EFA) and NVIDIA GPUDirect Storage (GDS), enabling up to 15x higher throughput to GPUs. This unlocks new possibilities for deep learning, autonomous vehicles, and HPC workloads. With EFA and GDS, Amazon FSx for Lustre can now achieve up to 1500 Gbps throughput per client, compared to 100 Gbps with the previous version. This is particularly exciting for training deep learning models, where large datasets and high GPU demands can slow down the process. With these enhancements, organizations can fully utilize the power of their GPUs and accelerate their workloads.