Pygwa is a general purpose data-parallel computing library. It is intended to support various data parallel programming, mostly GPGPU frameworks (AMD CAL in its initial versions). Pygwa features a two-layer design. Lower layer will be a binding to a particular framework/library, upper layer will provide a set of abstractions useful for solving various data parallel tasks.
Ubuntu/AMD64 + ATI card are initially supported. If you like porting pygwa to another platform, let me know and send me patches - I'll be happy to merge it with main tree.
One of project objectives is to perform some research about parallel (and data parallel) problems and come up with a set of tools useful for solving problems in parallel way and learning about parallelism in general.
View full history Series and milestones
trunk series is the current focus of development.