Re: The ROR bug - seems it wasn't a bug after all.

From: didier derny <*didier_at_aida.org*> Date: Mon, 12 Feb 2024 11:25:18 +0100 Message-ID: <8e65eb7d-7d0e-4ed2-90bc-d2350ab1c4ba_at_aida.org> · didier_at_aida.org

Frankly I don't know why it's a bug or not
The essential is that finally they made a great processor that powered our machines...

I am more concerned by the w65c51 and the w65c816 still sold knowing that the w65c51 is unusable and the w65c816 crippled by bugs In emulation mode

Le 12 févr. 2024, 08:39, à 08:39, Francesco Messineo <francesco.messineo_at_gmail.com> a écrit:
>On Mon, Feb 12, 2024 at 8:31 AM Francesco Messineo
><francesco.messineo_at_gmail.com> wrote:
>>
>> On Mon, Feb 12, 2024 at 8:12 AM Julian Perry
><jp_at_digitaltapestries.com> wrote:
>> >
>> > I'm sure I'm very late to the party, but have learned that the
>infamous ROR bug in 1975 6502's actually was not a bug - just an
>unimplemented feature.
>> > Seems like the builder of the Monster 6502,Eric Schlaepfer spoke
>with Bill Mensch in 1991 on the topic, and Bill confirmed that the ROR
>instruction was simply not implemented in the silicon at all - and was
>only added after demand for it became compelling. In a short youtube
>video he puts a compelling case for this - (including
>reverse-engineering, die-shots and circut: diagrams and I think is the
>definitive word on the subject.
>> >
>>
>> it was definitely a bug, no other manufacturer would add an
>> instruction on the second revision of a CPU. They couldn't fix the
>> design in time for the planned first release and that was a brilliant
>> move to not delay the CPU time to market but obviously they intended
>> to fix the missing instruction.
>> Both Chuck Peddle and Bill Mensch in their oral history confirmed
>that
>> the ROR instruction was planned but something went wrong during the
>> first masks layout and they better left it out instead of delaying
>the
>> 6502 release.
>
>and just to add some other facts: $66 $76 $6E $7E opcodes line
>perfectly into the ASL/LSR/ROL decoding mask and there're almost no
>holes in that line of opcodes, so they had left just the right holes
>in the decoding ROM but without planning for it? Give me a break.
>
>>