Maybe it's finally time I try out this debugger in VirtualT I've heard
mentioned a few times before.
Maybe I can step through both a 100 version and a k85 version and see
where they diverge as they run?
On 5/25/26 17:30, Brian K. White wrote:
Interesting, thank you.
I will study that.
I now have a single consolidated assembly source that produces both
RAM100 and RAM200 from the same source, with all machine differences
abstracted out to a heaser file for each machine, and an
overcomplicated makefile to build them.
That new shared version doesn't have 4-bank support yet, it just
reproduces the original binaries exactly so I can keep doing "make
verify" to catch when I break it while monkeying around with the source.
But now that I have both 100 and 200 disassembled (more or less, a lot
of the labels are still meaningless), I'm ready to add the 4-bank in
as an option so it will build both legacy and new versions.
And then maybe kc85 if I can actually find all the addresses.
https://github.com/bkw777/NODE_DATAPAC/tree/main/software/RAMDSK/src
On 5/25/26 05:33, B 9 wrote:
Oh, I just noticed the “BANK0” is an ASCII char, not an integer. So,
the full ALT code would look more like this:
|LDA BankNumMSG ; read current bank number (ascii) INR A ; add 1 SUI
BANK0 ; convert ascii 'n' to n ANI (NBANKS-1) ; is it > highest valid
bank number? MOV B,A ; copy bank num to B MOV C,A ; copy bank num to
C MVI A,PORT_CTL0 ; start with control port for bank 0 @LoopBN: JZ
@DoneBN ; ANI above or DCR below set Z, jump to commit ADI
BANK_CTL_OFFSET ; add 4 to control port number DCR B ; decrement B
JMP @LoopBN ; loop @DoneBN: STA CtlPort ; commit new control port
number MOV A,C ; get bank number ADI BANK0 ; convert back to ascii
character STA BankNumMSG ; commit new bank number display (ascii) |
—b9
On Sun, May 24, 2026 at 11:48 PM B 9 <[email protected]
<mailto:[email protected]>> wrote:
I’m no assembly expert, but at first glance at seeing the code, it
seems odd to me that you’re allowing BANK0 to equal 1 sometimes in
the follow stanza:
|LDA BankNumMSG ; read current bank number (ascii) INR A ; add 1 CPI
BANK0+NBANKS ; is it > highest valid bank number? JC
@UpdateBankNum ; BankNum <= highest, jump to commit MVI A,BANK0 ;
BankNum > highest, reset to '0' @UpdateBankNum: STA BankNumMSG ;
commit new bank number display (ascii) |
If you stay with 0-based counting and NBANKS=4, I believe you can
simply |AND 0b0011| instead of doing the compare and jump. Something
like this:
|LDA BankNumMSG ; read current bank number (ascii) INR A ; add 1 AND
NBANKS-1 ; Wrap back to 0 if it is > highest valid bank number STA
BankNumMSG ; commit new bank number display (ascii) |
Of course, even if I’m right, your original code is more flexible.
For example, my shortcut would break if you wanted to compile for a
situation where NBANKS is not a power of two. (I don’t know anything
about RAM100, but I noticed that the two values for NBANKS mentioned
were “2” and “4”).
—b9
On Sat, May 23, 2026 at 3:31 PM Brian K. White <[email protected]
<mailto:[email protected]>> wrote:
I've been working on RAM100.CO <http://RAM100.CO> to make usable
assembly source from a
disassembly so I can add support for more than 2 banks and maybe
port to
kc85, EU, make a common 100/200 source, etc.
I've gotten the 100 version working and added the support for 4
banks.
So now for 100 & 102, all 1M of MiniNDP is usable for files.
(not just 102 because I have an untested dip-40 pcb for 100 &
kc85)
I haven't started on 200 yet.
2 questions:
Is there a less convoluted way to do the ALT version of this
chunk of
code highlighted below? I feel like there must be a more direct
way to
arrange that loop without a jump at both the beginning and end.
Am I being silly even caring about it?
Both versions work fine and there's no speed or size problem.
I've scavenged enough bytes elsewhere that even the larger
version
still comes out actually one byte shorter than the original
binary
before padding just to make the bootstrap basic code with hard-
coded top
& end values for the old binary also work on the new binary.
https://github.com/bkw777/NODE_DATAPAC/blob/main/software/
RAMDSK/RAM100/4-bank/RAM100.S85#L929-L993 <https://github.com/
bkw777/NODE_DATAPAC/blob/main/software/RAMDSK/RAM100/4-bank/
RAM100.S85#L929-L993>
What the chunk does is increment or roll-over 2 values.
display bank number ascii '0' to '3'
control port number 129 to 141 step 4
One defensible functional difference I just realized is with the
alt
version I could have a hot key or call target for basic to jump
directly
to any arbitrary bank rather than stepping through a loop.
-- bkw
--
bkw
