2019-01557 - PhD Position F/M Automatic parallelization of real-time application onto multi-cores
Le descriptif de l’offre ci-dessous est en Anglais

Type de contrat : CDD de la fonction publique

Niveau de diplôme exigé : Bac + 5 ou équivalent

Fonction : Doctorant

Contexte et atouts du poste

This PhD is funded by the ES3CAP ("Embedded Smart Safe Secure Computing Autonomous Platform") collaborative project. The project aims to propose a software environment for multi-/many-core platforms, facilitating the transition from current architectures (distributed heterogenous or single-core/single-task) to industrially-viable centralized multi-/many-core solutions for high-performance, safe, and secure systems. The target market is that of new-generation critical systems (avionics, drones, autonomous vehicles).

In ES3CAP, Inria brings its expertise on parallel programming of critical software and on automatic allocation, real-time scheduling, and code generation for multi-/many-core platforms. Inria also brings its Lopht tool for automatic mapping and code generation, which has already been ported on the Kalray MPPA256 Bostan many-core. Lopht ensures by construction the functional correctness of the generated code and the respect of non-functional (real-time) requirements.

Mission confiée

In previous work [1,2] we have developed an automatic parallelization
method and tool for multi-period hard real-time applications that
targets clusters of the Kalray MPPA256 Bostan many-core [3], chosen
for its excellent support for high-performance, timing predictable
execution. Our method produces parallel code that is functionally
correct and satisfies by construction non-functional (real-time)
requirements without compromising efficiency (speed, memory
footprint). To achieve this, our method relies on a tight integration
between timing analysis, real-time scheduling, and code generation
around a precise timing model of the execution platform.

Our approach has two key points:

- The timing model, built upon a model of the memory system of the
  many-core allowing the precise and safe estimation of memory access
  interferences between parallel threads.
 
- Real-time scheduling and code generation choices allowing
  the synthesis of parallel code that is efficient and at the
  same time allows scalable and precise timing analysis.

The objective of this PhD is to extend the existing memory model and
parallelization method to cover other shared memory multi-core
platforms, such as the new-generation Kalray MPPA Coolidge, Power
T1042, ARMv8 Cortex A53, or Infineon Aurix TC27x. Each of these
architectures poses different problems: heterogenous memory (mix of
SRAM, DDRAM, and FLASH with different access times), NUMA memory
(access times depend on the distance between core and RAM bank),
on-chip interconnect topology, etc. On traditional Power and ARM
multi-cores the architecture description is only partially known, so
timing guarantees can only be partial.

[1] K. Didier, D. Potop-Butucaru, G. Iooss, A. Cohen, J. Souyris,
    P. Baufreton, A. Graillat. Efficient parallelization of
    large-scale hard real-time applications. Inria Research Report
    9180 https://hal.inria.fr/hal-01810176v2/document
[2] T. Carle, D. Potop-Butucaru. Predicate-aware, makespan-preserving
    software pipelining of scheduling tables. ACM TACO vol. 11(1),
    Feb. 2014.
[3] https://www.kalrayinc.com/portfolio/processors/

Principales activités

State of the art.

Research.

Dissemination through paper writing and interaction with industrial and research partners.

Tool development.

Compétences

We are looking for a candidate with an excellent record in computer science. Backgroud in the following fields is considered a plus: real-time systems, parallelization, distributed computing, compilation, programming languages, semantics, formal methods, computer architecture.

Avantages

  • 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 (after 6 months of employment) 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