[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Floating-point formats and standards

This page is part of the web mail archives of SRFI 58 from before July 7th, 2015. The new archives for SRFI 58 contain all messages, not just those from before July 7th, 2015.

Regarding my notes on IEEE 754 and its draft revision,
Shiro Kawai wrote:
> That's very informative summary.


> Just a note.  Currently, 16-bit binary floating-point numbers (called
> half floats) getting wider use in images; graphics processors directly
> deals with them, and ILM's OpenEXR image format stores values in them.

Ah, I didn't know that. Thanks!

> It may be too early to consider them in this srfi, but I can imagine
> I'd like to have uniform arrays of binary16 some day.

I suspect that's why 754R has added them to their draft. (There's a note
that they debated whether it should have a 4-bit or 5-bit exponent, so
the format might not be consistent from system to system yet.)
Bradd W. Szonye