a visit to Frys

[David Hough]

I recently bought a 20 MHz 386/SX for my kids to play PC games, and even though
it doesn't really need one, I decided to buy a floating-point processor for it.
I went down to Frys and discovered that the good, the bad, and the ugly are
all for sale next to each other, physically adjacent and financially adjacent
at $70-$80, shrink wrapped in cardboard and plastic just like high-fi 
accessories, all claiming compatibility with IEEE 754.

I did not expect it would come to this so soon,
neither ten years ago when David Stevenson
called P754 together at the Decathlon Club to celebrate passing the draft up
to the next level of IEEE standardization, nor twenty years ago when W. Kahan
remarked in the context of one of his CS 246 lectures (which I was writing
up the notes for) that some people wanted him to just sit down and prescribe
(rather than describe) how computer arithmetic should work, but he was 
reluctant to oblige them because he felt that once he started it would turn
into an infinite task, involving as it does hardware, operating systems,
languages, compilers, and libraries.

I finally chose a Cyrix for $74 rather than IIT, Intel, or ULSI.