Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn how to assess the performance of AI network and storage infrastructure critical for successful deployment of complex AI training and inference engines without requiring expensive GPU hardware investments. Explore methodologies and considerations for performance assessment across different topologies, examining host and network side considerations and key metrics. Discover the performance aspects of NICs/SmartNICs, storage offload processing, switches and interconnects while understanding benchmarking approaches for AI collective communications with RoCE transport. Analyze the impact on training convergence time and network utilization, including operational aspects of commercial networks such as proxies, encapsulations, connection scale and encryption effects on AI training and inference. Master performance considerations in AI fabric and storage deployment, optimization metrics for AI training and inference, and understand the role of various AI infrastructure elements that impact overall system performance.
