"We are very pleased to help offer a solution that allows HPC, AI and high-performance data analytics (HPDA) to work in the same infrastructure and eliminate time consuming data transfer for our users", stated Gary Tyreman, president and CEO of Univa. "Grid Engine has been deployed thousands of times in business-critical clusters. We want to ensure this community can add mixed HPC, AI and HPDA workloads to their environments in an efficient manner."
A converged environment for HPC and AI eliminates the burden of data transfer between multiple environments to enable faster time to insights. Through the workload-optimized performance delivered by Intel Select Solutions, users can reap the benefits of a converged environment that offers a 4x faster inferencing performance improvement with the built-in Intel Deep Learning Boost for a faster time-to-results. This solution gives HPC customers a path to adopt artificial intelligence on the 2nd generation Intel Xeon Scalable processor environment.
"AI-enhanced high-performance computing accelerates discovery and innovation for economic competitiveness and technology leadership", stated Trish Damkroger, vice president and general manager of the Extreme Computing Organization at Intel. "In collaboration with innovators like Univa, Intel Select Solutions for HPC and AI Converged Clusters give users the assurance they need to quickly deploy and get the performance results they want and expect, whether deployed on-premise or in the Cloud."
Univa Grid Engine manages workloads automatically, maximizes shared resources and accelerates the execution of any container, application or service. The solution can be deployed in any technology environment: on-premise, hybrid Cloud or Cloud-native HPC. By using Univa Grid Engine, enterprises and organisations can deliver products and results faster, more efficiently and with lower overall costs.
Additionally, workloads running on Univa Grid Engine are efficiently shared across machines in a data centre to optimize the use of the computing infrastructure. Scheduling policies can be applied to all work submitted to the cluster, ensuring high-priority jobs are completed on time, while simultaneously maintaining maximum utilization of all cluster machines. Along with Navops Launch, Univa Grid Engine can dynamically schedule workloads to a hybrid Cloud environment.