Re: [Chicken-users] thoughts on alternate "posits" / "unums" instead of traditional floats?

[Peter Bex]

On Tue, Jul 09, 2019 at 12:25:12PM -0500, Daniel Ortmann wrote:
> I am probably the last to run across this alternate floating point
> format ... but now I am curious.
> 
> If these posits were implemented in Chicken, what sort of work would be
> required?
> Would they replace the traditional floats?

It sounds like they would replace traditional floats, yes.

> Or would they be an optional part of the numeric stack?

They could be, but Scheme does not really allow for multiple types
of inexact numbers.  It would also slow everything down even further
due to even more dispatching.  Right now, if we know something's a
float/inexact, we can simply inline calls.

We rely quite a lot on the C compiler to do flonum code.

> https://www.nextplatform.com/2019/07/08/new-approach-could-sink-floating-point-computation/

From this article:

   Better yet, he claims the new format is a “drop-in replacement” for
   standard floats, with no changes needed to an application’s source code.

So it sounds like a modified C compiler and/or FPU would be the way to go
for this.  CHICKEN can then transparently make use of that via the C type
"double".

Cheers,
Peter

signature.asc
Description: PGP signature