hi

z80 test error report bug in that test

http://www.cpc-power.com/index.php?page=detail&onglet=dumps&num=12883

SCF and CCF instructions

I think this a known variant issue according to MAMEtesters: https://mametesters.org/view.php?id=2701 did you request a followup there?

Yeah, we're emulating the NEC Z80, which in this case has that difference from the Zilog original.

Sorry it seem to be that issue

Yeah, we're in C++ land these days tho, so creating a derived type with different behavior is far easier than 99% of what goes into MAME...

TODO notes in z80.cpp: "This Z80 emulator assumes a ZiLOG NMOS model." (not NEC)

and regarding SCF/CCF flags: "However, people from the Speccy scene mention that SCF/CCF X/Y results are inconsistent and may be influenced by I and R registers." - so, is it wise to base NEC Z80 on SCF/CCF X/Y differences?

The z80.cpp notes are from like 1997. I would believe the newer results from the 2018 test program over them. Also, MAME was arcade-only forever, and arcade games almost always used the NEC D780, so it's not shocking we ended up modeling it instead of Zilog.

Those notes are old yes, more like 2008 or so.
I don't know of other NEC vs Zilog differences, but regarding SCF/CCF X/Y flags we definitely emulate the Zilog.

*edit* according to http://www.cpcwiki.eu/forum/technical-support/z80-defective-tests/ it fails on a real Zilog.

Maybe a lot of arcades just starts working if we implement the NEC variation? :-)

No known software other than this test program cares about this difference