Biography
Dr Beau Johnston is a Research Engineer at Spectral Compute and an Honorary Academic at the Australian National University. His research sits at the intersection of high-performance computing, compiler technology, runtime systems, programming languages, and heterogeneous computing.
Beau completed his PhD in 2019 at the Australian National University, where he investigated architecture-independent workload characterisation—developing tools and techniques to understand how scientific applications behave without relying on details of any particular processor or accelerator. His research explored how these characteristics could be used to predict application performance, improve scheduling decisions, and guide the design of future heterogeneous computing systems.
Over the past decade Beau has worked across academia, government laboratories, and industry, including Oak Ridge National Laboratory (USA), Virginia Tech, the University of Bristol, and now Spectral Compute. His work spans heterogeneous runtime systems, performance portability, workload characterisation, scheduling, compiler evaluation, multi-GPU execution, and scientific computing. He currently serves on the Khronos Group SYCL Advisory Panel and was part of the Oak Ridge National Laboratory team recognised with a 2024 R&D 100 Award for the IRIS heterogeneous runtime.
At Spectral Compute, Beau’s research focuses on evaluating and improving the SCALE compiler stack. His work involves benchmarking HPC and AI applications across NVIDIA and AMD accelerators, analysing compiler and runtime behaviour, developing new performance evaluation methodologies, and identifying opportunities to improve portability and performance across modern heterogeneous systems.
Although now based in industry, Beau maintains an active research program through ANU and is particularly interested in collaborating with students who want to bridge the gap between research and real-world software systems. Many projects involve working directly with modern compiler technology, large-scale GPU systems, and open research problems that have immediate practical impact.
Research Interests
- High-performance computing
- Heterogeneous CPU/GPU systems
- Compiler technology
- Runtime systems
- Performance portability
- Multi-GPU execution
- Programming languages (CUDA, HIP, SYCL, OpenCL)
- Workload characterisation
- Scientific and AI computing
- Performance analysis and benchmarking
Student Projects
I’m always interested in supervising Honours, Masters, and PhD students with interests in compiler technology, heterogeneous computing, and high-performance software systems.
Current project areas include:
- Performance portability across NVIDIA and AMD GPU architectures
- Compiler evaluation and optimisation using SCALE
- Benchmarking modern HPC and AI workloads
- Multi-GPU runtime systems and scheduling
- Profiling and workload characterisation
- Scientific computing on heterogeneous hardware
- Programming models including OpenMP, OpenACC, CUDA, HIP, SYCL and OpenCL
Many projects are undertaken in collaboration with Spectral Compute, providing students with opportunities to contribute to research that directly informs the development of production compiler technology while still producing publishable academic outcomes.
If you’re interested in working on compilers, runtimes, GPU programming, or performance engineering, feel free to get in touch.