floating point MAC, duh!

[ranton] Considering the vast improvements in DSP-type IP and chip
resources, I'm baffled as to why multiply-accumulate (MAC) is not
included in the average set of floating point operator cores and FPUs.
It just seems fundamental to me. Using heavily pipelined atomic
multipliers and adders makes for an awefully slow MAC. [rantoff] Those
wishing to impelement such a thing should look up Yong Dou's work on
the internet. Here's one link:
ce.et.tudelft.nl/publicationfiles/973_567_f43-dou.pdf