recentpopularlog in

mcherm : floatingpoint   6

New Approach Could Sink Floating Point Computation
There is ACTUALLY a credible alternative to IEEE 754 for floating point.
floatingpoint  programming  computerscience  via:HackerNews 
5 weeks ago by mcherm
Floating-Point Determinism | Random ASCII
A really good analysis of floating point weirdness focused around the question of what is necessary to make floating point calculations repeatable, even on a different architecture. And it's a lot more complex than I would have thought.
programming  via:reddit  floatingpoint 
july 2013 by mcherm
Nondeterministic Floating-Point Conversions in Java - Exploring Binary
A bug in Java's conversion of certain obscure floating point values. Interestingly, this bug interacts with the switch from normal to JIT operations.
java  bug  floatingpoint  via:reddit 
february 2011 by mcherm
Bertrand Meyer's technology blog » Blog Archive » Reflexivity, and other pillars of civilization
Bertrand Meyer explains why he chose to have NaN == NaN in Eiffel even though that violates IEEE754. He has some fairly good reasons.
math  programming  eiffel  languagedesign  via:python-list  floatingpoint 
april 2010 by mcherm
JavaRanch Big Moose Saloon: int to float 16777217
The smallest integer that cannot be represented as a float, and why.
java  math  floatingpoint 
july 2007 by mcherm

Copy this bookmark:

to read