Heterogeneous Design Approaches to the Convergence of HPC and AI

This presentation addresses the following topic(s):

• Machine learning and HPC: Each worthy of its own investment, or better together? (Alternately: Marriage for the ages, or heading for divorce?)
• Specialization vs. standardization: How do you maximize specific application performance without creating silos?

The convergence of AI and HPC is leading to the need to handle more diverse workloads in HPC systems. This increase in diversity of workloads coupled with the ever-expanding landscape of available accelerators can be a challenge for designing HPC systems. We will explore the benefits of designing HPC systems with heterogeneous accelerator technologies to make the marriage of AI and HPC faster and more efficient. We will take a practical look at what can be achieved today and emerging technologies that will drive the HPC designs of the future.