Home Original page

floating point in 2.0

Tim Peters tim.one at home.com
Sun Jun 10 01:31:25 EDT 2001 
[Tim]
> HW is key, because without near-universal HW support for a
> thing that thing will never become widely accepted.  HW vendors used
> to punt on denorms because of the expense, but despite the expense
> everyone plays along with denorms now, and not because of user demand
> but simply because it's required.

[Stuart L. Anderson]
> None of the SGI boxes I've seen support denorms.  Also, the IBM
> mainframes and the Classic Crays don't use IEEE 754 in any form.  And
> then, there are the Alpha boxes running in VAX mode.

Had you drawn up a listing 10 years ago (when Python's numerics were being
designed), the list would have been very much longer.  For example, I worked
at Kendall Square Research then, and we punted on denorms too -- but 754
conformance hadn't worked its way into govt contracts by then either.  For
the love of God, Stuart, Cray even came out with the world's lamest *Ada*
compiler, just to check off that box <wink>.  There are a lot fewer
companies designing FPUs now, but conformance to the letter of 754 is
near-universal among new designs.

still-python-wants-to-run-on-everything-ly y'rs  - tim
More information about the Python-list mailing list