Future extreme-scale supercomputers will be heterogeneous and lead to new and challenging demands for efficient numerical algorithms and parallel software libraries. The aim of NLAFET is to tackle these challenges and ultimately deliver new scalable numerical libraries for fundamental problems in numerical linear algebra, including the solution of dense and sparse systems of equations and eigenvalue problems. Achieving this requires a co-design effort including developing novel algorithms, exploration of advanced scheduling strategies and runtime systems, offline and online autotuning, as well as avoiding communication and synchronization bottlenecks.
The positions will focus on design, implementation, and evaluation of parallel algorithms for matrix kernels and eigenvalue problems with regards to the extreme-scale challenges. The research will be done in close collaboration with recognized researchers and PhD students in the research group as well as with other partners in NLAFET.
Deadline for applications is April 4th, 2016.
For complete information of the positions see:
Postdoc (English): PDF or https://umu.mynetworkglobal.com/en/what:job/jobID:93318/
PhD (English): PDF or https://umu.mynetworkglobal.com/en/what:job/jobID:93348/
]]>You will be working in collaboration with world leaders in the field through an EU Horizon 2020 research grant. The NLAFET project focuses on delivering new algorithms and technologies for dense and sparse linear algebra on next generation supercomputers. The work at STFC will focus mainly on the strong scaling of sparse direct solvers using task-
based paradigms, though contributions will be made across the full range of the project. Impact will be delivered through publication of technical reports and research papers, through presentations at workshops and conferences and through the development and release of high quality research software.
Responsibilities will include:
About You:
The qualities that we are seeking can be grouped as:
Deadline: March 20th, 2016
More information can be found here: PDF
]]>