[fuse-emulator-devel] SCF and CCF instructions

[Stuart B. <sd...@nt...>]

Hi,

I've written a test program for SCF and CCF.  Woody has run it on a real
48K and a real +2A, and the results are quite surprising.  The test
program can be downloaded from:

 * http://homepage.ntlworld.com/wholehog/stuart/scfccf-dump.tap

The source is available from:

 * http://homepage.ntlworld.com/wholehog/stuart/scfccf-dump.asm

In February 2004, Fuse's implementation of SCF was changed so that it
copied bits 3 and 5 from A into F, instead of ORing bits 3 and 5 with A
with the original values of bits 3 and 5 of F.  The original discussion
for this is here:

 * http://sourceforge.net/mailarchive/forum.php?thread_name=20040217082529.GH17981%40philos.lan.philosys.de&forum_name=fuse-emulator-devel

The test results from the +2A that Woody tested seem to match Fuse's
previous behaviour.  I believe that the Z80 in this +2A is a Zilog Z80.
These results are available at:

 * http://homepage.ntlworld.com/wholehog/stuart/woody+2a.txt

On the 48K, bit 3 seemed to be unaffected by the either operation, but
bit 5 appears to be obtained by ANDing the original value of bit 5 with
something else, but I'm not sure what.  Here are the results:

 * http://homepage.ntlworld.com/wholehog/stuart/woody48k.txt

As Fuse's current behaviour was verified by Ian Greenway on a real
Spectrum, there must then be at least *three* different behaviours for
SCF on real hardware!
-- 
Stuart Brady