limonce@pilot.njin.net (Tom Limoncelli) (12/28/88)
I recently was testing a possibly defective A501 RAM/Clock card [damn
Christmas returns, I hate 'em] for a customer's A500. I noticed that
different amounts of RAM were free on each boot-up. I was always
booting on the same disk so I thought that it might be the card. The
difference was usually about 30-150 bytes or so, and the systest
program showed no problems, [and I did a lot of swapping with the
store's equipment and the customer's equipment] and I concluded that
the card was OK.
I did find that the amount of free-RAM on an A500 is different
depending on how you boot it. I haven't tested this out on my B2000
at home yet. Here is what I found:
2 different variables. Each test performed with only one change in
variable.
Variable 1 - machine state before boot: OFF (the machine was
off) or REBOOT (the machine was on, and was rebooted).
Variable 2 - did I wait for the "WorkBench" request: SCREEN
(yes, the screen was displayed) or NOSCR (the disk was inserted before
the A500 had a chance to display the request)
All tests performed on a A500 with A501 card. Kickstart 1.2 and
WorkBench 1.2 33.61. The disk was a virgin AmigaDOS disk (as far as I
know :-). The test was ran twice with different results. A lot of
customers started coming in so I had to stop experimenting. Later I
tried repeating a particular combination (didn't record the results)
and got varying answers like I did here.
First Run Second Run
Var-1 Var-2 Free Memory Free Memory
OFF SCREEN 926288 926320
REBOOT SCREEN 926288 926344
OFF NOSCR 926408
REBOOT NOSCR 926384
REBOOT NOSCR 926408
" " 926384
" " 926384
" " 926384
So, it looks like the code that displays the friendly hand with a
workbench disk isn't freeing all the memory it could. You can
reliably get 96 extra bytes by not displaying that screen. It also
looks like there is a random 24 bytes that do or don't appear
depending on the phase of the moon. :-)
Now I'm not the person that complains about 120 missing bytes, but I
thought that this was an interesting point. Maybe I'm just wasting
bandwidth or maybe someone will research the code that does this and
find a tiny yet-undiscovered bug that, when fixed, will result in
fewer crashes, world peace, and better TV sitcoms. :-)
("Then Tom woke up and realized it was all a dream...") :-)
...or maybe this is a "known bug" slated for fixing in 1.4.
...or maybe I've been working too hard.
-Tom
--
Tom Limoncelli Drew University Madison NJ 201-408-5389
tlimonce@drunivac.Bitnet limonce@pilot.njin.net
"Fences make good neighbors" -Frost "I want an MMU" -Me
Standard disclaimer? No, we're still on the dpANS disclaimer.