A two-hour XSEDE tutorial at 12 p.m. MDT on Wednesday, November 3, will introduce participants to the development platform for NVIDIA A100 and A40 GPUs. Twenty percent of the computing power for Derecho, NCAR’s next supercomputer, will come from NVIDIA A100s. 

The NVIDIA HPC software development toolkit (SDK) is a suite of tools for programming NVIDIA GPUs for high-performance computing, artificial intelligence, and machine learning. The SDK includes a full set of compilers for C, C++, and Fortran, with GPU-offload models such as OpenACC, OpenMP, and CUDA. The SDK also includes a suite of GPU-accelerated math libraries and communication libraries. For AI, NVIDIA provides containers with fully optimized implementations of popular machine learning frameworks for both training and inference. NVIDIA will also present on the A100 and A40 GPUs and describe what hardware features users can take advantage of.

To attend, register by November 2 through the XSEDE User Portal. Space is limited to the first 300 registrants.