Two Proposals

Ryan Newton rrnewton at gmail.com
Tue Oct 4 15:50:25 CEST 2011


Just anecdotally I remember we had this problem with Accelerate.

Back when we were using it last Spring for some reason we were forced by the
API to at least nominally go through lists on our way to the GPU -- which we
sorely hoped were deforested!  At times (and somewhat unpredictably), we'd
be faced enormous execution times and memory footprints as the runtime tried
to create gigantic lists for feeding to Accelerate.

Other than that -- I like having a nice literal syntax for other types.  But
I'm not sure that I construct literals for Sets and IntMaps often enough to
profit much...

  -Ryan


On Tue, Oct 4, 2011 at 9:38 AM, Roman Leshchinskiy <rl at cse.unsw.edu.au>wrote:

> George Giorgidze wrote:
> >
> > This extension could also be used for giving data-parallel array literals
> > instead of the special syntax used currently.
>
> Unfortunately, it couldn't. DPH array literals don't (and can't really) go
> through lists.
>
> In general, if we are going to overload list literals then forcing the
> desugaring to always go through lists seems wrong to me. There are plenty
> of data structures where that might result in a significant performance
> hit.
>
> Roman
>
>
>
>
> _______________________________________________
> Glasgow-haskell-users mailing list
> Glasgow-haskell-users at haskell.org
> http://www.haskell.org/mailman/listinfo/glasgow-haskell-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.haskell.org/pipermail/glasgow-haskell-users/attachments/20111004/260ac48c/attachment.htm>


More information about the Glasgow-haskell-users mailing list