Computation Offloading Between Real-Time and General-Purpose Operating Systems (exjobb)
The opportunity:
We are looking for a master thesis student to join our R&D department and our multinational organization with exciting opportunities to work with advanced technologies. Our team is dedicated to creating a future where innovative digital technology enables cleaner and more reliable energy systems. If this opportunity excites you, we encourage you to apply!
Modern multicore processors provide significant computing power, but efficiently utilizing this performance in real-time systems introduces new challenges. This thesis will investigate computation offloading between a Real-Time Operating System (RTOS) and a General-Purpose Operating System (GPOS) running as separate virtual machines in a real-time hypervisor environment.
The project will explore which computational workloads can benefit from offloading, considering the trade-off between increased computing performance and the communication overhead introduced when transferring tasks and data between operating systems. Particular attention will be given to latency, timing determinism, computational acceleration, and deadline compliance in real-time applications.
How you’ll make an impact:
- Develop and benchmark representative computational algorithms, including GEMM, SpMV, Stencil and FFT, in sequential and parallel implementations.
- Investigate mechanisms for transferring computational tasks and data between the RTOS and GPOS environments.
- Develop a C++ middleware solution for computation offloading using Hitachi Energy's RTLink message-passing library.
- Benchmark computational acceleration against communication and data-transfer overhead.
- Evaluate key real-time characteristics including latency, timing determinism and deadline compliance
- Analyze which types and sizes of computational workloads benefit most from RTOS-to-GPOS offloading.
- Explore advanced opportunities such as prediction of deadline misses, degraded or partial computational results, and potentially OpenCL-based compute kernels.
- Provide recommendations for effectively applying computation offloading in future real-time systems.
- Live Hitachi Energy’s core values of safety and integrity, taking responsibility for your actions while caring for colleagues and the business.
Your background:
- Bachelor’s or Master’s studies in Computer Science, Computer Engineering, Electrical Engineering, Embedded Systems, or a related field.
- Good programming skills, particularly in C++.
- Interest in embedded systems, real-time systems, multicore processing, virtualization or parallel computing.
- Understanding of operating systems and computer architecture is beneficial.
- Experience with benchmarking, profiling, or performance analysis is a plus.
- Knowledge of RTOS, hypervisors, OpenCL, or heterogeneous computing is considered an advantage, but is not required.
- Proficiency in English, both written and spoken.
More about us:
Are you ready for an exciting new challenge? Does the description above sound like you?
Applications will be reviewed on an ongoing basis, so don’t delay, apply today!
Recruiting and preliminary host manager: Lucas Bottura
Thesis supervisor: Bruno Chianca