[CaCL] parallelizable computations: broken tool

[Nathan Rasmussen]

I mentioned there is a library that lets you write routines in (a
parallelizable subset of) Python, compile them via C++, and run them on
multiple cores or on GPUs. I also mentioned that it was broken by some
recent updates to gcc. As of three weeks ago, it compiled but didn't link
right.

Back when it worked, this is how it worked:
http://devblogs.nvidia.com/parallelforall/copperhead-data-parallel-python/

I have no expertise in compilation/linking issues, so unfortunately I'm not
qualified to fix this thing. The last time I wrote in a compiled language
was in 1997, and the language was (a beginner's subset of) Pascal. But I
thought I'd send this out in case any of you with better qualifications are
interested.

Here's the most recent version: https://github.com/bryancatanzaro/copperhead

If you do decide to take it up, let me know and I'll introduce you to the
author. He's smart and generous, just too busy to fix it himself right now.

Nathan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osu.edu/pipermail/cacl/attachments/20141008/0a965aca/attachment.html>