Workshop: LLVM-HPC2026: The Twelfth Workshop on the LLVM Compiler Infrastructure in HPC
Authors: Dominik Adamski, Sergio Afonso Fumero, Akash Banerjee, Pranav Bhandarkar, and Andrew Gozillon (Advanced Micro Devices (AMD)); Michael Klemm (Advanced Micro Devices (AMD), OpenMP Architecture Review Board); and Michael Kruse, Jan Leyonberg, Dan Palermo, Krzysztof Parzyszek, Hafiz Abid Qadeer, and Krishna Chaitanya Sankisa (Advanced Micro Devices (AMD))
Abstract: Contemporary supercomputers are massively parallel, heterogeneous systems with accelerators---mostly graphics processing units (GPUs)---that provide additional compute and memory performance. Fortran continues to have a large developer base with new code written every day. The OpenMP Application Programming Interface (API) thus constitutes one of the key programming languages to support developers in their need to write portable, high-performance code for such heterogeneous systems. In this paper, we will highlight the evolution of the AMD Fortran Compiler (AMD Flang) in the past twelve months. As AMD Flang is based on LLVM Flang, much of the evolution is also relevant for the LLVM project via pull requests that have been contributed or will be contributed in the future.
Back to LLVM-HPC2026: The Twelfth Workshop on the LLVM Compiler Infrastructure in HPC Archive Listing
Back to Full Workshop Archive Listing