This Silverton Consulting White Paper lifts the fog surrounding deep learning storage I/O requirements for training and inferencing. Analyst Ray Lucchesi outlines the key considerations driving deep learning I/O and finds such workloads don’t need the implausible performance levels that organizations have been led to believe.
“Although DL training and inferencing are well understood these days, the IO activity or bandwidth required to perform DL suffers from many misconceptions.”
Ray Lucchesi, President, Silverton Consulting
