The ultimate benchmark

M.ROBINS mlr at hounx.UUCP
Wed Oct 9 12:20:34 AEST 1985


       Below is	the procedure for the ultimate benchmark:

       The #1 question of the day is, of course, how fast  can	the
       various	computing  systems  we	all  deal  with	perform	out
       favorite	applications: nothing?	We have	all devised various
       means   of   benchmarking   this	  most	 popular   computer
       application, with various degrees of success.

       The time	is ripe	to remove  this	 experimentation  from	the
       realm of	pure research in order to apply	it to everyday use.
       It is clear from	our tests that most computers we deal  with
       can do nothing very quickly, in fact, it	is feasible for	the
       computer	to  do	less  in  seconds  than	 experts  do  in  a
       lifetime.

       It has been pointed  out	 to  me,  however,  that  computer-
       iterated	  nothing  is  really  something,  and	that's	not
       nothing.	 Everyone knows	that  you  can't  get  nothing	for
       something.  What	 is  clearly needed is a way to	do a lot of
       nothing without the overhead of iterating,  which  tends	 to
       detract	from the amount	of nothing performed; the iterating
       being something rather than nothing.

       A novel approach	is to remove all  power	 from  the  system,
       which  removes most system overhead so that resources can be
       fully  devoted  to  doing  nothing.   Benchmarks	  on   this
       technique  are  promising; tremendous amounts of	nothing	can
       be produced in this manner,   Certain  hardware	limitations
       can limit the speed of this method, especially in the larger
       systems which require  a	 more  involved	 &  less  efficient
       power-down sequence.

       An alternate approach is	to pull	the main  breaker  for	the
       building,  which	 seems to provide even more nothing, but in
       truth has bugs in it, since it usually inhibits the  systems
       which  keep the beer cool.  The best approach seems to be to
       provide a specialized piece of hardware for  removing  power
       solely  from  the  computer system itself, without affecting
       local beer-cooling and cheese-dip-mixing	equipment.  I  have
       found  this system to be	most satisfactory.
       Let me demonstra %(#^



More information about the Comp.unix mailing list