PhD Position F/M Programming real-time applications on scratchpad-based multicore architectures

The  Inria University of Lille centre, created in 2008, employs 360 people  including 305 scientists in 16 research teams. Recognised for its strong  involvement in the socio-economic development of the Hauts-De-France  region, the Inria University of Lille centre pursues a close  relationship with large companies and SMEs. By promoting synergies  between researchers and industrialists, Inria participates in the  transfer of skills and expertise in digital technologies and provides  access to the best European and international research for the benefit  of innovation and companies, particularly in the region.

For more  than 10 years, the Inria University of Lille centre has been located at  the heart of Lille's university and scientific ecosystem, as well as at  the heart of Frenchtech, with a technology showroom based on Avenue de  Bretagne in Lille, on the EuraTechnologies site of economic excellence  dedicated to information and communication technologies (ICT).

The PhD student will be part of the SyCoMoRES team of Inria Lille & CRIStAL lab, which currently hosts 6 fellow PhD students. Lille is a city close to Brussels, Paris & London, easily reachable by train, with a large student population and a number of cultural places & events.

PhD students are appointed for a duration of 3 years. We plan to organize weekly research meetings with the PhD student.

An Embedded Real-Time System (ERTS) controls a physical device in its environment, at a rate adapted to the device and its environment. ERTS can be found in many industrial domains, including automotive, avionics, nuclear plants or automated production lines. They are often of critical nature; a malfunction may damage properties or even put human lives in danger. Developers of ERTS must ensure not only that the system computes correct values, but also computes values at the right time.

Multicore hardware platforms have now become the norm for ERTS, due to their potential for supporting complex applications. The memory hierarchy comprises multiple levels of caches, one or more shared interconnects, and a central shared random-access memory (DRAM). Tasks running on different cores may simultaneously access shared components of the memory hierarchy, causing contention and interference that can delay their execution. As a consequence, performing a timing analysis on such a platform is a difficult problem, because this requires to finely analyse the evolution of the cache content, the scheduling of tasks, and the contention resolution mechanisms.

A scratchpad memory is a local, directly addressable fast memory whose content is explicitly managed by the programmer. Scratchpads are an attractive alternative to caches for ERTS, because they provide better control over when memory transfers occur, thus simplifying timing analysis. However, this requires to manage memory transfers directly in software, which can be tedious and error-prone for large scale applications.

The objective of the thesis is to contribute to a framework for the development of ERTS on multicore hardware architecture where core-local memories are scratchpad memories. Our goal is to automatize memory allocation and data-transfers between the different memory subsystems, so as to ease the development of ERTS on such platforms.

First, we would like to design an Application Programming Interface (API) at the Operating System level, consisting of simple primitives
that abstract the underlying platform. This could for instance consist in extending an existing Real-Time Operating System, such as FreeRTOS for instance.

Second, we would like to explore the compilation of ERTS for such a hardware platform. The system will be programmed with a high-level data-flow language, such as Prelude for instance, which was specifically designed for the development of critical ERTS. The compiler with automatically generate the allocation of code and data in the different memory subsystems, as well as the memory transfer instructions.

The following activities will be carried out during the thesis:

• Bibliographic research
• Proposing solutions for the identified research directions
• Writing research papers on the thesis results
• Teaching (optional)

Technical skills: a good background on computer science and embedded systems.

Languages: English. French is not required.

Other valued, appreciated: compilation, real-time systems, hardware architecture

• Subsidized meals
• Partial reimbursement of public transport costs
• Leave: 7 weeks of annual leave + 10 extra days off due to RTT (statutory reduction in working hours) + possibility of exceptional leave (sick children, moving home, etc.)
• Possibility of teleworking and flexible organization of working hours
• Professional equipment available (videoconferencing, loan of computer equipment, etc.)
• Social, cultural and sports events and activities
• Access to vocational training
• Social security coverage

2,300 euros (Monthly gross salary)