8

The first production runs of the MOS 6502, in mid-1975, had a bug in the ROR instruction that caused MOS to "remove" it from the instruction set by omitting it in the intial documentation. An updated version with a working ROR was released no later than May 1976.

While the Apple 1 was first demonstrated in late 1975, Apple Computer Company wasn't formed until March 1976, according to this interview with Woz. (Actually, April 1st, according to Wikipedia) National Geographic claims the release date of the Apple 1 to be April 11th, 1976, but I don't know where they got this date from. Wikipedia has another source saying it went on sale in July 1976, which, since the parts were ordered no more than a month before delivery of the assembled computers (according to both the first two references above), seems easily to leave enough time for their initial set of assembled computers to have used the fixed 6502.

So did any Apple 1s ship with a 6502 without ROR? Possibly some were assembled by others (who bought just the PCB) with a no-ROR CPU; do we have or know of any examples of this?

As far as software goes, unsurprisingly neither the original Woz monitor nor Apple 1 BASIC have any ROR instructions in them; these were developed well before the Apple 1 was ever sold as a product. But was there any other Apple 1 software sold or widely shared that deliberately worked around the ROR bug?

Improve this question

3 Answers 3

Reset to default
4

In this video from an auction service, the 6502 fitted is clearly shown (eg. at 2:05) as a white-ceramic packaged example with a 1576 datecode, corresponding to early April. This is before the ROR bug is certain to have been fixed, so Apple may have received a batch from near the end of the pre-fix production run.

It is consistent with reliable estimates of when the Apple 1 production run occurred, so may well be the original CPU supplied with the machine. Other ICs are clearly visible with consistent datecodes, eg. 7618, 7602, and from '75. Four 74161s are visible in one shot with identical 7611 datecodes, while the newest codes I could see are on a pair of the Signetics shift registers, 7620.

Week 20 corresponds to mid-May, which is about right for a suggested on-sale date for the completed Apple 1 of sometime in June, given that it took Woz and Jobs almost a month to order and then assemble all the required components, after obtaining funding and credit. They succeeded in obtaining payment in time to pay their component suppliers on 30-day terms.

But the CPU is roughly a calendar month older than the Signetics chips.

Improve this answer
1
  • 1
    I think that supply chains were much longer in the 70's than nowadays where we are obsessed with just-in-time (just-too-late?) kanban systems. It could take a while for stuff to get to the end user, particularly when it is a couple of people in a garage.
    – Jon Custer
    Commented Mar 3, 2020 at 14:11
3

For the second part:

But was there any other software sold or widely shared that deliberately worked around the ROR bug?

The most prominent may be Microchess, originally developed for the KIM but ported to next to any 6502 system and many non-6502, like TRS-80 and CP/M. It might be on of only a few (the only one I know) commercially sold applications for the Apple 1.

While it wasn't released until late 1976, Peter Jennings had to develop it around that bug, as there was quite a number of KIMs delivered with pre-correction 6502. Still today he tends to bring up the topic when talking about Microchess (or 6502 in general as a comment by scruss may imply).

Improve this answer
9
  • 1
    Peter Jennings told me that he replaced the 6502 in the CHM Apple 1 with an age-appropriate ceramic chip with all the original bugs. I'm guessing it was missing ROR. I'll ask him next time I see him.
    – scruss
    Commented Mar 2, 2020 at 14:38
  • 1
    @cjs As usual you're way to fast to dismiss information. Microchess was as well available for the Apple 1 (like many other systems) back in the days. It was eventually one of the very few commercial packages (the only I know of) that were ever adapted to the Apple 1.
    – Raffzahn
    Commented Mar 2, 2020 at 18:56
  • 2
    I suggest you update your answer to clarify it, then; saying "Microchess (for the KIM)" with no mention at all of the Apple 1 implies can reasonably be read as "it was for the KIM only."
    – cjs
    Commented Mar 2, 2020 at 23:52
  • 1
    @scruss Oh, he did. One time he loaded Microchess onto a KIM of mine and the first thing he did was checking for the bug - just to tell me that I got 'the real thing' and mentioning how he had to work around.
    – Raffzahn
    Commented Mar 3, 2020 at 0:24
  • 2
    No, I think the foremost function of an SE site is to give the OP the information he wants, which often enough is not done by merely giving answers to exactly and only what was asked or, worse yet, declaring the question "off topic" and deleting it. (This seems to be the core of most of our disagreements.)
    – cjs
    Commented Mar 3, 2020 at 9:05
1

On the AppleFritter forum I happened to ask "Did the original Apple 1s built by Apple have the ROR bug?" and Cory Cohen ("Cory986") replied:

Yes they did. Only the last boards out of Apple had a fixed CPU. This is one reason Mike Willegal has to make a special version of his memory test software which would run on original Apple-1 boards.

I have no way to judge the validity of this myself, but I am told by Mike Willegal that, though this was not the reason for the memory test software change, Cory Cohen would know. Willegal also says he has vague memories about people with original Apple 1s encountering problems due to the the ROR bug, and so leans towards at least some (presumably non-hobbyist-built) Apple 1s having it.

(This blog entry gives the reason for the ROR fix in Willegal's memory test program: someone who had built a Mimeo found a CPU old enough to have the bug, and used it in his replica.)

Improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .