PhD Position F/M Data placement in heterogeneous memory levels
Plus aucune candidature n'est acceptée pour cette offre d'emploi
Inria
il y a 2 mois
Date de publicationil y a 2 mois
S/O
Niveau d'expérienceS/O
Temps pleinType de contrat
Temps pleinDonnées / Big dataCatégorie d'emploi
Données / Big dataContexte et atouts du poste
Position Overview:
This role represents a collaborative opportunity between two Inria teams: CAMUS and STORM.
It is related to the NUMPEX PEPR Exasoft project ( https://numpex.org/fr/ ) and will allow interacting with the partners of the project.
Location:
The selected candidate will be positioned within the CAMUS team, based in Illkirch Graffenstaden, within the ICube laboratory.
Team Focus:
The research conducted in this position will aim to bridge the gap between task-based runtime system, high-performance computing (HPC) programming and its applications.
Mission confiée
Task-based programming has proven to be very effective in implementing high-performance applications on heterogeneous architectures. With this method, programmers describe their algorithms as tasks and their dependencies. Additionally, programmers inform the runtime system about the data being used. As a result, the runtime system can move the data across different memory nodes. Moreover, the scheduler (an internal component of the runtime system) decides where the various tasks will be executed, and the data will be moved accordingly.
The objective of the project is to imagine and implement advanced strategies to decide where the data should be moved when there is a choice. These strategies will have to predict where the data will be used (which is related to the scheduling policies) and when they will be used. It will require a deep knowledge of the hardware bus/network and latency/throughput.
Key aspects of the research include:
This research project is part of the national initiative PEPR NUMPEX, the French Exascale program that aims at designing and developing the software components that will equip future exascale machines. NumPEx will deliver Exascale-grade numerical methods, software, and training, allowing France to remain a leader in the field. It will contribute to bridging the gap between cutting-edge software development and application domains to prepare major scientific and industrial application codes to fully exploit the capabilities of these machines. Application domains of the NumPEx program include, but are not limited to, weather forecasting and climate, aeronautics, automotive, astrophysics, high energy physics, material science, energy production and management, biology, and health.
Principales activités
Main activities:
Compétences
Technical Skills and Proficiency Required:
Programming Languages:
Interpersonal Skills:
Additional Valued Qualities:
The applicant must have a MS degree or equivalent.
Avantages
Rémunération
2100 € gross/month the 1st year
Position Overview:
This role represents a collaborative opportunity between two Inria teams: CAMUS and STORM.
It is related to the NUMPEX PEPR Exasoft project ( https://numpex.org/fr/ ) and will allow interacting with the partners of the project.
Location:
The selected candidate will be positioned within the CAMUS team, based in Illkirch Graffenstaden, within the ICube laboratory.
Team Focus:
- CAMUS Team: Specializes in compilation and high-performance computing. The team currently comprises 4 postdoctoral researchers/engineers, 9 PhD students, and 8 permanent researchers.
The research conducted in this position will aim to bridge the gap between task-based runtime system, high-performance computing (HPC) programming and its applications.
Mission confiée
Task-based programming has proven to be very effective in implementing high-performance applications on heterogeneous architectures. With this method, programmers describe their algorithms as tasks and their dependencies. Additionally, programmers inform the runtime system about the data being used. As a result, the runtime system can move the data across different memory nodes. Moreover, the scheduler (an internal component of the runtime system) decides where the various tasks will be executed, and the data will be moved accordingly.
The objective of the project is to imagine and implement advanced strategies to decide where the data should be moved when there is a choice. These strategies will have to predict where the data will be used (which is related to the scheduling policies) and when they will be used. It will require a deep knowledge of the hardware bus/network and latency/throughput.
Key aspects of the research include:
- Create semi-automatic strategies: In the first step, we will create strategies that will be potentially guided by the users.
- Create automatic strategies: In the second step, we will improve the strategies to make them autonomous.
- Benchmark on different architectures: We will continuously run benchmarks to measure and quantify the improvement of our contributions.
This research project is part of the national initiative PEPR NUMPEX, the French Exascale program that aims at designing and developing the software components that will equip future exascale machines. NumPEx will deliver Exascale-grade numerical methods, software, and training, allowing France to remain a leader in the field. It will contribute to bridging the gap between cutting-edge software development and application domains to prepare major scientific and industrial application codes to fully exploit the capabilities of these machines. Application domains of the NumPEx program include, but are not limited to, weather forecasting and climate, aeronautics, automotive, astrophysics, high energy physics, material science, energy production and management, biology, and health.
Principales activités
Main activities:
- Imagine new methods to decide where the data should be placed/moved
- Implement high performance code
- Run benchmark
- Participate to conference and publish the work done
Compétences
Technical Skills and Proficiency Required:
- Knowledge of parallelization approaches
- Proficient in CUDA programming and experienced in implementing high-performance computing (HPC) applications.
- Familiarity with HPC environments and compilation processes.
Programming Languages:
- C (mandatory), C++
- CUDA
Interpersonal Skills:
- Capable of collaborative teamwork as well as independent work.
Additional Valued Qualities:
- Creativity.
The applicant must have a MS degree or equivalent.
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
Rémunération
2100 € gross/month the 1st year
RÉSUMÉ DE L' OFFRE
PhD Position F/M Data placement in heterogeneous memory levelsInria
Illkirch-Graffenstaden
il y a 2 mois
S/O
Temps plein