30 Aug
2007
30 Aug
'07
10:18 p.m.
On 8/31/07, Dan Piponi
Right. But the functions and data that we are trying to map and fold could be anything, so we are required to have the full functionality of Haskell running on the GPU - unless the compiler can smartly figure out what should run on the GPU and what shouldn't. All in all, this could be a fairly ambitious project.
Well, yes, I didnt say it would be easy ;-) It could be useful to make a ballpark estimate of what kind of performance we'd get running Haskell on a GPU, compared to running on a CPU. Any ideas on how to do this? By the way, is there some reason we couldnt give the entire haskell program to every thread, and simply pass in something roughly equivalent to the program counter in with the data?