[Numeric-interest] Underflow signalling definition

[Chris Hinds]

I am interested in arguments for 'before rounding' and 'after rounding'
determination of tininess, and where the 754R spec may choose to go on
this.  The spec continues to specify the options as did the 754 spec,
but there has been talk in the past regarding removing some of these
options.  I know that this is an open issue, and am asking for
preferences and arguments.

 

I have read the paper by Cuyt, Kuterna, Verdonk and Verschaeren (revised
version) and appreciate the issues.  

 

Many thanks!

 

Chris Hinds