Scaling NumPy Applications from 1 CPU to Thousands of GPUs - IPAM at UCLA

Explore the potential of scaling NumPy applications from a single CPU to thousands of GPUs in this 37-minute conference talk presented by Seshu Yamajala from SLAC National Accelerator Laboratory. Discover how cuNumeric, a drop-in replacement for NumPy, enables scientific applications to overcome performance bottlenecks and handle larger datasets by scaling to multiple nodes and leveraging GPU acceleration. Gain insights into the advantages of Python and NumPy for developing complex scientific applications, and learn how to extend their capabilities beyond single CPU execution. Recorded on May 4, 2023, as part of IPAM's workshop on Complex Scientific Workflows at Extreme Computational Scales at UCLA, this presentation offers valuable knowledge for researchers and developers working with large-scale scientific computations.

Seshu Yamajala - Scaling NumPy applications from 1 CPU to thousands of GPUs - IPAM at UCLA