🚀 go-pugleaf

In article <80@aimt.UUCP| breck@aimt.UUCP (Robert Breckinridge Beatie) writes:
|In article <1178@csib.UUCP>, jwhitnel@csib.UUCP (Jerry Whitnell) writes:
|
|>             286/287     M68020
|>             16 Mhz     16 MhZ
|> Scientific   0.14        0.43
|> Commerical   0.48        1.44
|> Systems      0.70        1.50
|> General      0.31        0.93
|> 
|
|I notice that the 286 column is labeled as 286/287.  Does this mean that
|the 286 also had a numeric co-processor?  Since the 68020 column isn't
|labeled 68020/68881 does that mean that the 020 did NOT have a numeric
|co-processor?  If so, then this is amazing.  The 020 without floating point
|support beats a 286 with floating point support at SCIENTIFIC applications
|and does it by better than 3-1!!!  This seems too good to be true.  Could
|some kind soul post either confirmation or rufutation of my assumptions.  I
|don't have easy access to the article in question.

According to the instruction sequences given in the article, the 80286
did have  a 287, whereas floating point for the 020 says "estimated from
Motorola software".  The 020 is actually slower doing floating point
then the 287 (big surpise :-)), but the instruction mixes choosen are
such that floatting point does not have a strong influence.  For the
Scientific mix, it is < 15%, for the others, all less the 3%.  

|
|Even if the 020 did have a numeric co-processor, it looks as if, "yes Virginia
|the 020 does beat the pants of the 286."
|-- 
|Breck Beatie
|uunet!aimt!breck


Jerry Whitnell                           It's a damn poor mind that can only
Communication Solutions, Inc.            think of one way to spell a word.
						-- Andrew Jackson