[llvm-dev] Open Position - Compilation for Heterogenous Systems (FPGA, Math. Optimization, Polyhedral Compilation) (original) (raw)
Tobias Grosser via llvm-dev llvm-dev at lists.llvm.org
Wed Aug 9 05:44:53 PDT 2017
- Previous message: [llvm-dev] ind variable
- Next message: [llvm-dev] [ThinLTO] Suggestions on how to traverse SCCs in the Module Summary Index bottom up
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Dear all,
we are glad to announce four open positions at Polly Labs (hosted at ENS Paris) and at ETH Zurich to work with us towards developing scalable compilation technology for heterogeneous systems (including FPGAs) and large-scale real-world programs.
Starting off from the solid foundations the LLVM Polly loop optimizer and accelerator compiler provides, we aim to scale the underlying techniques towards larger programs, better (or optimal) transformations, and modern programming languages (Julia, Javascript, ...). For this journey we will combine linear and non-linear solvers, integer polyhedra based program generation techniques, JIT compiler technology, automatic tuning, accelerator compilation, ... and likely many other ideas. Emerging platforms such as FPGAs are one of our core focus points. All of our research results will be published as open source! Depending on the position, you might work in close collaboration with one of our industrial partners.
We are looking for excellent candidates for (PRE)DOCTORAL and POST-DOCTORAL positions!
If you are (or know somebody who is):
- Expert in compilation and code generation (e.g., LLVM, gcc, …)
- Strong mathematical background (e.g., linear programming, (non)-linear optimization, …)
- Interested in doing theoretically profound research applied in practice and published as open source
with additionally experience in:
- Polyhedral loop modeling (e.g., Polly, isl, Pluto)
- Accelerators (e.g., CUDA, OpenCL, Vulkan)
- Development high-quality software and interactions with open source communities
and for the FPGA positions experience in:
- High-level synthesis (e.g., Xilinx Vivado, SDAccel)
- Overlay networks
get in touch!
Our job offers are pretty demanding. Still, we do not hire by ticking boxes, but look for exceptional candidates. If you are able to write a non-linear branch-and-bound solver over night, but never heard of GPUs, please talk with us. Similarly, if you can program FPGAs in and out, but need a crash course in polyhedra, get in touch!
Detailed job descriptions: http://grosser.es/jobs.html
Best, Tobias
- Previous message: [llvm-dev] ind variable
- Next message: [llvm-dev] [ThinLTO] Suggestions on how to traverse SCCs in the Module Summary Index bottom up
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]