Re: R6502AP

On Sunday 30 November 2014, 14:36:07 Gerrit Heitsch 
<gerrit@laosinh.s.bawue.de> wrote:
> On 11/30/2014 02:25 PM, silverdr@wfmh.org.pl wrote:
> > On 2014-11-30 02:48, Segher Boessenkool wrote:
> >>> I ran the tests. Tried earlier but they take much longer than I thought
> >>> so I finished only tonight. And.. the MOS 6502AD passes all the tests
> >>> from that suite. The R6502AP fails repeatedly on one called "pc64-aneb".
> >>> Since the mnemonics are different from what I once learned I don't know
> >>> what command this is. Probably something from the AND family. But the
> >>> difference is there every time I run the tests (why - BTW - does it have
> >>> to spin the motor all the time?)
> >> 
> >> That seems (from the name) to be testing ANE, a.k.a. XAA, op $8b.
> >> 
> >> It is the most unstable unsupported op, giving different results
> >> _per run_ on some MOS/CSG devices as well.  It is heavily temperature,
> >> process, batch, phase-of-the-moon, you-name-it dependent :-)
> > 
> > Mayby I should retest it during full moon :-)
> > 
> > But since we only passed new moon phase - currently all runs are
> > consistent: all three MOS chips I have left pass, while all Rockwells
> > fail right away. It may be of little practical meaning but it seems to
> > show that there is some difference in both chip's implementations.
> 
> Doesn't have to the implementation. It could be enough that they were
> made in 2 different factories.
> 
> But remember, Commodore used the R6502AP in their disk drives without
> any problems. So unless you plan something VERY esoteric, a failure in a
> test of a known unstable illegal opcode means nothing if all other tests
> pass.

these two opcodes do not give the same results on various C64s either for that 
matter - which is why i updated the lorenz suite to check for the magic values 
first.

-- 

http://www.hitmen-console.org    http://magicdisk.untergrund.net
http://www.pokefinder.org        http://ar.pokefinder.org

I can make it crash!



       Message was sent through the cbm-hackers mailing list