🚀 go-pugleaf

The SI rating is somewhat biased.  What you should do is use the MIPS,
Landmark, or CPUSPD benchmarks.  The SI can be altered with simple things such
as wait states, ram fresh cycle, bumping up your ram to the highest speed that
speed of RAM.  Just by running SPEEDER (a ram refresh altering utility), I
bump my SI up by .3 from 15.6 to 15.9 on my 16 MHz/1 wait AT.  SI is a general
performance benchmark, not a CPU specific bench.  You can use SI for your hard
drive also, not very informative, but it's there.  The moral of the story is
that you want a CPU only bench, not a system bench (which is what SI is).  You
can bump your SI up by using a 16ms hard drive, but does that affect the
execution speed of programs?  No.  Similiar, on my system with an XT
controller, the total SI is 11.8.  That's a performance reduction by 1/3
almost just by adding hard drive I/O to the bench.  So interleaving, DMA
speed, and all of those other goodies can affect an SI.  Don't let an SI fool
you.  It's not that big of a deal, those tricks that are used to get that
extra 1.0 of SI you can probably do yourself by reinterleaving your hard
drive, altering ram refresh cycles, optimizing your hard drive regularly with
a disk optimizer, etc.  But NEVER use an SI as a basis for how fast the CPU
is.  
 
I can best summarize this message with a quote:
 
 "Advertising is legalized lying."  - H.G. Wells

 /*--------------------------------------------------------------------------*
  * Flames: /dev/null (on my Minix partition)
  *--------------------------------------------------------------------------*
  * ARPA  : crash!pnet01!jca@nosc.mil
  * INET  : jca@pnet01.cts.com
  * UUCP  : {nosc ucsd hplabs!hd-sdd}!crash!pnet01!jca
  *--------------------------------------------------------------------------*/