alan@pdn.UUCP (Alan Lovejoy) (08/24/87)
The September issue of BYTE has yet another 68020/80386 benchmark
article, in which they admit to some errors in the previous results
and list new, improved results.
More importantly, they include numbers for the DSI '020 board (that
fits in AT's). This baby is faster than everybody else ('020 or '386)
with the exception of Dhrystone, where some of the '386 boxes are
slightly faster. Of course, they're still using compiled code,
so this still proves nothing.
Their main argument in their defense is that they're benchmarking
production systems available to the "average user", not CPUs.
Obviously, they did not succeed in communicating this to their
readers, probably because the headlines for these articles spoke of
'386 vs '020 and not Mac II vs. Model 80 or Hypercharger vs Deskpro 386.
Also, they should seriously consider the true needs of their readers,
which include other things than simply whose box is currently faster.
Readers need to know whose box/architecture has the greater potential.
After all, when you buy a system, you also are buying into a CPU
architecture (you probably want your next system to have a CPU
compatible with your current one).
--Alan "MOVE.L (A0)+,(A1)+" Lovejoy