[Novalug] [OT] The "real low level" of computers (it's not what you think) -- WAS: Packages v. ports v. source

Clif Flynt clif at cflynt.com
Tue Mar 30 04:59:19 EDT 2010


Bryan: Thanks for a nice, concise history of processor development.

Back-inna-day, I learned computer hardware/software on a PDP-8.  This
edition of the 8 was built of quad logic-Nor chips and Dual Type D
flipflops.  No other chips need apply.

The assembler was so simple, you could hand assemble your code in
your head - this bit means I/O, this bit means write, next word is
address, etc.

At that point, it made good sense to learn assembler to understand
the hardware.  We could toggle a loop on the front panel, watch der
blinkenlights, and see the data move on a 'scope.

In the '80s, I'd write in C and rework the important parts in hand
assembler.  (I once got a 10% improvement on my 68k machine by
rewriting the integer-multiply subroutine and patching all the
binaries.)

Nowadays, I write in Tcl and rework the compute intensive parts in C.

I agree that understanding assembler is unnecessary for someone
working at the application level.

For someone who expects to work at the driver/kernel level, I think at
least some time with assembler on a 68000, PDP-11, or similar "clean"
architecture is useful, just to understand how things behave in a
simpler world.

In a similar vein, an auto mechanic should tear apart and rebuild an
old slant-6 engine with a carbuerator, distributor and 2 valves per
cylinder, just to understand how engines work, and then start worrying
about 8 valve, fuel injected, computer controlled engines.

Clif

-- 
... Clif Flynt ... http://www.cwflynt.com ... clif at cflynt.com ...
.. Tcl/Tk: A Developer's Guide (2nd edition) - Morgan Kauffman ..
.. 17'th Annual Tcl/Tk Conference:  2010,   Oak Brook, IL  USA ..
.............  http://www.tcl.tk/community/tcl2010/  ............








More information about the Novalug mailing list