[NLRS] 439MHZ Radar

[Baker, Donn B]

Ahhh, maybe.  
If you look JUST at "clock speed," you can get that impression.  What you
have to measure, though, is "work."  That is, how many clock cycles does it
take to execute an instruction, and how much useful work does that
instruction do ?  Today's systems complicate that with techniques such as
"pipelining," where multiple instructions are in different phases of their
execution at each clock cycle.  

There are also two general "types" of computers: complex instruction set, or
CISC; and reduced instruction set, RISC.  RISC was a "big thing" a few years
ago, but has sort of fallen out of favor.  RISC is based on the idea that a
few, simple but very fast-executing instructions are better than many,
slower-executing instructions (CISC).

For example, a RISC machine may have ADD, STORE, and COMPLEMENT but not
DIVIDE, or MULTIPLY.  The last two functions can be accomplished with
sequences of ADD, STORE, and COMPLEMENT.  RISC machines are nice and easy to
understand, but in practice may take longer to do something useful that a
CISC machine.

Today, serious comparisons of computing systems are done by "benchmarks,"
not the comparison of clock speed, work capacity, or MIPS (millions of
instructions per second).  Benchmarks are established and agreed upon by
groups such as the Transaction Processing Council,  and the results can be
compared between different systems and manufacturers.  (By the way, the
current definition for 'MIPS' is "meaningless indicator of processor speed
!")

"Desktop PCs" and mainframes serve very different purposes, and really can't
be compared (or at least shouldn't).  Mainframes are lousy word processors,
and can't play DOOM for all the tea in China.  But even a 200GHz Pentium XXX
desktop can't support the NASDAQ stock exchange (1.25 billion shares traded
per day), or support 10,000 banking transactions per second.  Besides, would
you want YOUR bank account maintained by Windows ?  I sure as heck don't!

(Seriously, Microsoft wants to play in the "mainframe world" very badly.
Dataserver and even XP are steps in that direction.  It took the mainframe
people 25 years to get to the levels of reliability and performance they run
today... it will take MS the same.)

Several years ago, it was very fashionable to call mainframes "dinosaurs,"
implying that they were dying out.  In the real world, dinosaurs ruled the
earth for 170 million years. 

73 Donn
WA2VOI/0

-----Original Message-----

SNIP

 The most recent desktop
PCs have made a few year old mainframe computer look slow and puny.

73, Jerry, K0CQ