Cet émulateur est également le plus complet des outils de développements X68000 existant aujourd’hui. Il s’agit d’un fork de l’émulateur XM6 v2.05 de PI avec une pléthore de nouvelles fonctionnalités. Les changements sont pour la plupart liés à l’interface utilisateur et un accent a été mis sur le développement et les fonctions de débogage plutôt que sur l’exactitude de l’émulation, cependant il y a suffisamment d’améliorations pour qu’il soit recommandé d’utiliser cette version plutôt que XM6 v2.06 finale.
Les changements:
– Scaling for drive numbers (in the status bar panes) was improved but
only for newer systems, which are expected to have the necessary font.
– Made some minor improvements to several of the processor-related subwindows.
For example, in the « Modify » submenu, you can now type ‘0’ instead of « D0 ».
– When editing multi-word instructions by manually entering hexadecimal values,
the remaining words are no longer NOP-padded. For example, if you were to
change a conditional branch with a word-sized displacement value into an
unconditional branch (by changing the first word to 6000), the displacement
will now remain as-is.
– There is now a memory hashing function integrated into memory windows.
It calculates a standard CRC-32. The starting address is preloaded from
the current memory window location but it can be changed. Next you specify
the (exclusive) ending address or the size in bytes. The result is displayed
on the status bar and also copied to the clipboard. As an example, one could
use this feature to check if a block of memory has changed over time. In order
to facilitate such usage, the CRC is also written to the log.
Bug fixes:
– Several of the most recent releases had some negative performance impacts.
– Fixed an unlikely but easily reproducible crash.
– Fixed what I believe is a minor bug in the cycle counting bookkeeping.
– This change is unlikely to have a noticeable effect.
– Fixed a rare* crash when loading state.
* The bug had been around since the original XM6 without anyone reporting it.
Specific circumstances were required in order to trigger it, and even then
it wasn’t guaranteed to happen. But it was easy to duplicate if you knew how.
– Fixed a delightful state-loading bug that had no effect on casual users.
– A rarely* used instruction was not being emulated. (* We know this empirically.)
– The disassembler was decoding certain invalid instructions inconsistently.
