Accelerating MLPerf v0.6 Deep Learning Performance on Red Hat OpenShift
Diane Feddema, Principal Software Engineer, Red Hat Inc. & Mehnaz Mahbub Cluster Systems Engineer, Supermicro
About this talk
Containerized ML Applications on OpenShift provide many benefits to users in terms of optimizing setup, managing software stacks and visualizing whole system performance. In this webinar, Supermicro and Red Hat discuss the performance benchmarks which measured performance for MLPerf with Openshift on Supermicro GPU servers.
In the performance measurement, Supermicro and Red Hat implemented grafana dashboards backed by the Prometheus time series database to monitor performance from a whole-system perspective. Find out more about how this measurement and leave the session with an understanding of how to evaluate whole-system performance for ML applications and how to use OpenShift to run and observe ML workloads.