News for hackers. Stuff that matters.

A comment on Unum Computing: An Energy Efficient and Massively Parallel Approach to Numerics. Parent comment

Re:Unum Computing: An Energy Efficient and Massively Paralle... by dnautics on Friday July 24, 2015 @06:52PM (#9945725)
I think the correct answer is (0,inf) which is a value you can get by dropping to a lower resolution within the same unum environment.
[ Parent | on HN ]
Re:Unum Computing: An Energy Efficient and Massively Paralle... by tromp on Friday July 24, 2015 @07:07PM (#9945775)
Ah, that makes sense. The 4-bit representations do not suffice by themselves, but imply the support of some or all of the smaller representations as well, and I guess you always need a representation for [-inf,inf].
[ Parent | on HN ]