Data Parallel Haskell
Stc
Date: 2009-10-22
Time: 11:45
Room: BBL room 471
Speaker: Erik Hesselink
Title: Data Parallel Haskell
Abstract
Concurrent and parallel programming has become a hot topic, as
multicore machines have become mainstream. Haskell is a promising
starting point in this area, because of its purity and type system.
These enable safe concurrency by default, and allows large program
transformations. Data Parallel Haskell aims to provide the programmer
with a high level interface of parallel arrays, giving performance
competitive with hand-written C. In this talk, I will talk about the
programmer's interface to Data Parallel Haskell, the implementation
(which uses advanced Haskell techniques like type families and generic
programming), performance and the current state of the project.
