May 26, 2022 – The University of Bristol’s High Performance Computing (HPC) Research Group has established an Intel oneAPI Center of Excellence, building on its history of working closely with Intel in the areas of performance portability and advanced computing architectures. The center develops best practices for achieving exascale performance portability using open cross-architecture programming oneAPI and Khronos Group’s SYCL abstraction layer for cross-platform heterogeneous programming to ensure that scientific codes on massive heterogeneous supercomputing systems have a high can achieve performance.
The University of Bristol has had a long and successful working relationship with Intel, which led to the establishment of the UK’s first Intel Parallel Computing Center in 2014. This new Intel oneAPI Center of Excellence expands the joint venture to enable high-performance applications on more processors today and in the future.
Performance portability at Exascale with SYCL and oneAPI
Scientific codes strive to achieve high performance on heterogeneous systems composed of a combination of CPU and GPU architectures. These applications need to run efficiently across a range of processors and take full advantage of the capabilities of the hardware on which they run. Achieving portability is a challenge since different programming methods are typically used for different architectures. The collaboration between Bristol and Intel through this oneAPI Center of Excellence will develop and share best practices on how to make application performance portable on exascale through unified heterogeneous programming using SYCL together with oneAPI.
Through the Bristol Intel oneAPI Center of Excellence, Bristol will develop, document and share gold-standard best practices for developing scalable SYCL applications for exascale. Because SYCL is supported by Intel’s oneAPI DPC++/C++ compiler and code migration tools, developers can write portable programs for heterogeneous systems that run with high efficiency on a range of processor classes. By presenting the best practices for realizing power portability, this project is a critical part of highlighting the benefits of open-standard, heterogeneous parallel programming models for achieving exascale milestones and providing an open choice of hardware by facilitating the adoption of new cross-vendor architectures .
The center will initially work with two applications Bristol has contributed to the new SPEC HPC benchmark suite: CloverLeaf and TeaLeaf. SPEC HPC is a forward-thinking suite of applications capable of stressing heterogeneous systems with input sizes ranging from single nodes to exascale-sized problems. These codes are representative of important scientific applications, so it is critical that these algorithms perform well in the diverse exascale processor landscape. Bristol will develop SYCL scalable versions of these key benchmarks to demonstrate best practices for performance portability, including optimizations for Intel and non-Intel CPUs and GPUs.
dr Tom Deakin, Principal Investigator and Lecturer in Advanced Computing Systems at Bristol, said: “SYCL’s parallel programming model allows for power portability, but in practice this is not always easy to achieve. Becoming a oneAPI center of excellence will allow us to share our expert knowledge on SYCL and performance portability, and enable developers to more productively write fast HPC applications on CPUs and GPUs.”
Professor Simon McIntosh-Smith, Head of the HPC Research Group in Bristol, said: “This new partnership builds on our longstanding collaboration with Intel. Our research group is a global leader in finding solutions that make it possible to transfer scientific code to the most advanced hardware, and we see oneAPI as an important piece of this puzzle.”
“This new Center of Excellence at the University of Bristol builds on a longstanding collaboration with Intel that includes applied research on performance portability, standards and best practices for developers of high-performance applications,” said Joe Curley, vice president and general manager of Intel Software Products and Ecosystem Group . “We look forward to discovering and sharing new developer best practices based on multi-architecture, multi-vendor languages like SYCL as part of the oneAPI initiative.”
About High Performance Computing at the University of Bristol
The University of Bristol has a vibrant HPC community with around 1,700 users of the supercomputing services hosted by its Advanced Computing Research Centre. The university is ranked 8th in the UK for computer science by the Guardian University Guide 2021 and trains around 60 highly talented undergraduate computer science students each year in high-performance computing, including parallel programming languages.
OneAPI is an open, unified, and cross-architecture programming model for CPUs and accelerator architectures (GPUs, FPGAs, and others). The standards-based programming model simplifies software development and delivers uncompromised performance for accelerated computing without proprietary commitments, while allowing integration with existing code. With oneAPI, developers can choose the best accelerator architecture for the specific problem they are trying to solve without having to rewrite software for the next architecture and platform.
Source: University of Bristol