Quote:
Originally Posted by Thomas Richter
IEEE 754 is the best you get can get, with limited bits available per number. Thus, the problem is that you have unrealistic assumptions about "floating point numbers".
|
After watching the video seminar linked I think you will change your mind, getting more decimals from less number of bits with Posits compared to IEEE 754, and the added benefit of a bit telling you if the float was exact or not.
Quote:
Originally Posted by Thomas Richter
|
I will, but not sure what the point is? It's from 1991 (older than 2017, the date of the video seminar).