View Proposal


Proposer
Rob Stewart
Title
Targeting FPGAs with parallel functional languages
Goal
Develop a Verilog backed for the functional language Accelerate
Description
FPGAs are reconfigurable chips and offer promise of very high performance, low powered targets for accelerated computation. They have potential in many domains including High Performance Computing, Cloud Computing, embedded processing and autonomous robotics. FPGAs are usually programmed at very low levels with hardware description languages, and sometimes at the higher level C language. High level parallel array processing languages like APL, Accelerate and Chapel abstract above hardware, usually targeting multicore CPUs and GPUs. This project will involve writing an FPGA backend for the Accelerate DSL, an embedded language in Haskell, Backend options are OpenCL or C++, from which High Level Synthesis tools will generate FPGA hardware designs. This is a compiler research project that will establish how to produce high performance (outperforming parallel CPUs and GPUs) and efficient hardware from very high level array processing codes.
Resources
Background
Url
External Link
Difficulty Level
Challenging
Ethical Approval
None
Number Of Students
1
Supervisor
Rob Stewart
Keywords
Degrees