Hacker News Reader

It's really interesting hearing about the language designs were looking at floating-point hardware as an optional feature back then. These days I barely think about its usage, and I love seeing my perspective challenged on what comes "in the box".

View on HN

Floating point hardware was optional for a long time. 486 in 1989 was first Windows PC CPU with floating point standard. Before that it was an add on co-processor for the 386. New 386 models were coming out until 1992, so it took a long time before you could count on floating point support being there.

There are still a lot of embedded microprocessors without floating point support.

The 486sx was also in the mix for some time after that.

Which had the amusing feature that some motherboards let you add a '487 FPU' [0], which was actually a 486DX that took over completely from the SX.

[0] https://en.wikipedia.org/wiki/X87#80487

Even the relatively expensive Motorola 68k chips didn't have built-in floating point until the 68040 in 1990.

I'd forgotten it was that early. The Sun box on my desk was a 68020 with a 68881. Everything having to do w/ FP, certainly on our DSPs (which is silly), we had to implement a knock-off of IEEE 754 in software ourselves.

floating point hardware wasn't available on the majority of computers in mid 2009 : ARM.

Reply on news.ycombinator.com