ICL Research Profile
The High Performance Conjugate Gradients (HPCG) benchmark is designed to measure performance representative of modern scientific applications relying on discretizations of Partial Differential Equations (PDEs). It does so by exercising the computational and communication patterns commonly found in real science and engineering codes, often based on sparse iterative solvers with complex multi-level preconditioners. HPCG exhibits the same irregular accesses to the main memory and fine-grain recursive computations that dominate large-scale scientific workloads to simulate complex physical phenomena.
The HPCG 3.1 reference code was released in March 2019. In addition to bug fixes, this release positioned HPCG to represent modern PDE solvers better and made it easier to run HPCG on production supercomputing installations. The reference version is accompanied by multiple binary or source code releases from AMD, ARM, Intel, and NVIDIA, which are carefully optimized for these vendors’ respective hardware platforms. The current HPCG performance list was released at SC21 and features over 200 entries across the supercomputing landscape. TOP500.org has also tracked HPCG results since June 2017.
In Collaboration With
- Sandia National Laboratories
- National Nuclear Security Administration
- The United States Department of Energy