Retrotronik RAM card for HP 48GX

[rprosperi]

(07-02-2021 06:56 PM)Joe Horn Wrote:  

(07-01-2021 10:01 PM)Eric Rechlin Wrote:  4 MB cards mostly function but have a problem with the last port, so they generally aren't recommended.

Does anybody remember (or have a link to) what the Port 33 Bug was, exactly? I vaguely remember that Port 33 will actually exist (if you have a 4 MB card in slot 2 of a 48 GX), but that it misbehaves in some nasty way. Any info in that regard would be helpful. Thanks in advance! Related question: Is there any way of write-protecting ONLY Port 33, so as to avoid getting bitten by the Port 33 Bug?

Only a vague memory, not unlike yours, that Port-33 does in fact exist, but it is unreliable as data can be corrupted. I also kinda-remember that the 48GX uses one of the address/data lines used when Port-33 is active, for some other purpose, which is why it can clobber data you store there.

Since most port operations are controlled by the user (though I suppose an errant program could do that too) it's generally safe IIRC to simply ignore it. What isn't clear, since Port-33 is definitely there and exists if a 4MB card is installed, is what happens when you use wildcards to access data across any card, e.g. EVAL or PURGE ":&:MYDATA". Again if you don't store anything there, it should be OK... statistically speaking.

The RAM cards implement write-protect at the card level, so almost certainly not. There were programs/libraries that implemented port-locking, but I imagine such locking is only effective within the scope of that program. The FAQ has some fairly detailed notes from Dave Arnet on the various bits and signal lines involved; perhaps something could be done in assembler, but if nothing has emerged yet, it's not likely needed. Besides, all 11 people that own a 4MB RAM card know about the problem...