|start | find | index | login or register | edit|
by earl, 5867 days agoMark Jones: Type Classes with Functional Dependencies (2000) - "Type classes in Haskell allow programmers to define functions that can be used on a set of different types, with a potentially different implementation in each case. [..] A commonly requested extension to support `multiple parameters' allows a more general interpretation of classes as relations on types, and has many potentially useful applications. This paper illustrates [..] problems that can occur with multiple parameter type classes, and explains how they can be resolved by allowing programmers to specify explicit dependencies between the parameters."
For a short and quite accessible teaser what you may want to solve via multi-parameter type classes see the FunDeps page at the Haskell Wiki.
Quite related to yesterdays Cell-induced daydreaming is the [create Accelerator] work [via LtU@] (and similar GPGPU efforts in general). From the abstract of their paper:
"Accelerator [is] a system that simplifies the programming of GPUs for general-purpose uses. Accelerator provides a high-level data-parallel programming model as a library that is available from a conventional imperative programming language. The library translates the data-parallel operations on-the-fly to optimized GPU pixel shader code and API calls."
So even if IBM does not follow Hannibals day-dream and keeps Cell's SPEs as floating-point SIMDs, a similar approach to Accelerator would allow leverage in implementations of data-parallel languages:
Our [Accelerator] work demonstrates that it is possible to compile highlevel data-parallel language extensions to mass-market parallel processors that are currently available.
5 active users
|earl.strain.at • esa3 • online for 7628 days • c'est un vanilla site|