[comp.sys.amiga] Is the Amiga 1000 faster than the Amiga 500?

bryce@hoser.berkeley.edu (Bryce Nesbitt) (09/15/87)

I keep hearing speculation that says the Amiga 500 is slower than the Amiga
1000.  Nobody has any facts, any real evidence.  "I heard it from a
friend".  Perhaps that friend worked for Atari :-).  We'll see.

The speculationists gave figures from 2% to 25% slower.  I too was guilty
of this crime... but I said it was just as fast as the A1000.  I based this
on knowing that they could not have changed the clock, and that Agnus' DMA
timing was tight... how they could have messed with that was beyond me!

So, after another episode of A500 bashing I decided to put science on my
side.  I wrote a very simple wait loop.  I Disable() interrupts at the
start and don't Enable() them until after ten minute of processor-
intensive nothing.  Since the 68000 is the same and the computers are to
run identical graphics modes this is a valid test.

I do _not_ use the internal timer!  A long time ago I posted a friend's
TimeRAM to this net.  I know now that this program is defective, and can
not be relied upon to give accurate readings.  (Not the fault of the timer)
This is why I never released a summary... the times where bad and only
showed blatant figures such as the fact C Ltd's memory boards, and those in
the German A2000 are painfstead.  For you sticklers with
a *real* timer I provide a signal on an external port that can give you a
reading accurate to a couple of hundred nanoseconds.

----------
How to use "SpeedT" (see next posting for executable)

Copy "SpeedT" to a RAM disk of some sort.  Type "SpeedT" from the CLI.  (I
almost did an icon...).  I use it from the CLI so I can put the pointer in
the very top left corner.  The computer's power LED will immediately dim.
You will have about 2 seconds to ready your stopwatch.	The LED will go
bright and you had better start timing.  Now wait 9.5 minutes and come back
to your Amiga. Things should still be locked up.  Within a minute or so the
LED is going go dim again.  RESIST the temptation to hit your stopwatch!!
This is the 2 second warning.  When the LED goes bright, hit your button.

People I know that work as switchers at a TV station can hit a stopwatch
anime base.

I have also provided for those of you that have a non-human who can take a
timing with a digital signal.  At startup "Speed" blinks the LED, sets the
parallel port to output and writes: %10001100 to it.  After the first 2
seconds it blinks the LED and writes %10000011.  At the 2 second warning it
blinks the LED.  After the timing period it blinks the LED and sets the
porr that your timing device needs.

---------------------------------
The speeds for my Amiga 1000 are:

Standard V1.2 release Workbench.  Morerow'ed to 704*232 pixels resolution.
With Spirit 1.5 Meg internal  : 09:59:59
FAST memory enabled	      : 09:59:57
			      : 09:59:58
Spirit disabled (w/nofastmem) : 10:46:86
CHIP (or "slow") memory only  : 10:46:83
			      : 10:46:85

A500  the A1000 run at exactly the same speed.
						^^^^^^^
What memory board you buy can have great impact on that speed.
				   ^^^^^

One note: The A501 or compatible memory board for the A500 will be slower
than a quality FAST memory board for average use, and much slower for heavy
graphics use.  This is because the A501 can contend with CHIP memory.  With
two jumpers this can be located and _classified_ by the Operating System as
CHIP memory, eliminating the disadvantage.  It cannot actually be _used_ as
CHIP memory until Commodore-Amiga releases a new version of the Agnus chip
that has wide enough DMA registers.

Second note: I can't recommend the Spirit board until they add a 74F32
speed OR gate that takes AS* and their OVR* and produces a final OVR*
output.
---------------------------------

Send me your results.  I *will* summarize this time.  I need the resolution
of the screen that was showing, the computer, and the memory board that was
in use.  If you have more than one you must be sure that only the one under
test is active.  By default the Operating System takes from any additional
memory first; use a program to disable your extra memory to test the speed
of CHIP ram.  ("noFASTmem" or another program with the same purpose
probably came with your memory board)


|\ /|  . Ack! (NAK, EOT, SOH)
{O o} . 
 (")	bryce@hoser.berkeley.EDU -or- ucbvax!hoser!bryce
  U