HPC Software Performance Engineer W/M (BBP)
With EPFL - Blue Brain Project in Genève - CHMore jobs from EPFL - Blue Brain Project
Posted on December 06, 2018
About this job
Job type: Contract
Experience level: Mid-Level
Role: System Administrator
Industry: Academic Research
Company size: 5k-10k people
Company type: Public
c++, python, linux
Your mission :
The EPFL Blue Brain Project (BBP), situated on the Campus Biotech in Geneva, Switzerland, applies advanced neuroinformatics, data analytics, high-performance computing and simulation-based approaches to the challenge of understanding the structure and function of the mammalian brain in health and disease. The BBP provides the community with regular releases of data, models and tools to accelerate neuroscience discovery and clinical translation through open science and global collaboration.
We are currently looking for a: HPC Software Performance Engineer (W/M) to strengthen BBP simulation and High Performance Computing team.
Main duties and responsibilities include :
Participating to the development of scientific parallel high-performance applications targeting Exascale heterogeneous computing platforms.
Benchmarking & optimizing BBP HPC software on supercomputing architectures (POWER-GPU and Intel x86/KNL)
Prototyping computational kernels on next-generation interconnects and programmable hardware
Presenting results of developments and optimizations at various scientific venues
Literature survey of HPC software and hardware technologies Essential skills and experience required:
Professional experience in writing parallel scientific codes in modern C/C++ and Python using best software development practices including test driven development and continuous integration
Professional experience in distributed and concurrent programming (MPI, OpenMP, Pthreads, OpenACC, OpenCL, CUDA …)
Professional experience in developing and optimizing HPC software on accelerator systems (Nvidia GPU, Intel Xeon-Phi, …)
Professional experience in benchmarking, profiling, optimizing and debugging software on HPC systems with high level expertise in using HPC development software (Intel VTune, Scalasca, HPCToolkit, Allinea DDT/Map, Roguewave Totalview)
Experience using UNIX/Linux operating systems
Ability to solve problems and perform research independently
Good team player and fluent English in speech and writing Preferred:
Experience in building performance models
Experience in using low-level (assembly) and FPGA programming
Proven experience working on petascale parallel numerical applications
Experience writing high-throughput and evaluating parallel I/O code.
Experience in working in large code base projects with team of 5 developers or more
Your profile :
- Master’s degree or higher in High Performance Computing, computer science or related field with three to seven years of experience in a Supercomputing Center or equivalent.
- Successful track record in developing and optimizing software on petascale systems.
We offer :
- An internationally visible and rising project in simulation-based research in neuroscience using supercomputers
- A young, dynamic, inter-disciplinary, and international working environment