Encore une update mais importante cette fois, juger plut\u00f4t:<\/p>\n
Changes:
\n* CPUs with an implicit address shift (e.g., the TMS34010 shifts addresses
\nright 3 bits; the TMS32031 shifts addresses left 2 bits) now automatically
\nget that shift applied to addresses in their memory maps and in calls to
\ninstall_memory_read\/write_handler(). This eliminates many ugly macros in the
\nmemory maps. Affected CPU cores are the ADSP210x, CCPU, PIC16C5x, TMS3201x,
\nTMS3202x, TMS32031, and TMS340x0. Note that the CPU cores themselves still
\nneed to pass shifted values to the core read\/write routines. All drivers
\nusing these cores have had their icky macros and bitshifts removed.<\/p>\n
* Harvard architecture CPUs now have data memory fully implemented now.
\nAffected CPUs are the ADSP210x, TMS3201x, TMS3202x, PIC16C5x, and CCPU. All
\ncores have been modified to fully split program and data memory. All drivers
\nusing these cores now specify explicit program and data memory maps. This
\nalso gets rid of a ton of ugly macros used in memory maps.<\/p>\n
* The memory_set_bankhandler calls are now officially history. There were
\nfewer than a dozen drivers using them, and they all have been converted over
\nto using install_mem_read_handler instead where appropriate. This eliminates
\nthe last 8-bit-specific hack in the memory system. As a result, RAM reads
\nand writes out of banked memory should be slightly faster on 8-bit CPUs
\nbecause they are handled inline.<\/p>\n
* New dynamic memory allocation scheme. The previous method relied on
\nparsing the read\/write handlers. The new method simply looks at the
\ntop-level memory map after it has been populated, and allocates memory for
\nany mapped 16k chunk. This will likely cause more memory to be allocated
\nthan before, but the process is less error prone. Also, checks are now in to
\nmake sure that the memory_region for the program address space is the
\nminimum necessary to support non-sparse address spaces.<\/p>\n
* Fully integrated save\/restore memory blocks. Before, the save\/restore
\nsystem had its own memory map walking routines. That’s gone now. Instead,
\nmemory blocks are registered as they are allocated by the dynamic memory
\nallocator. Note that this will likely cause save states to be bigger than
\nthey were before (and certainly, they will not be compatible), but disk
\nspace is cheap, live with it.<\/p>\n
* Address spaces are now determined dynamically based on whether a given CPU
\nreturns a non-zero value for the address bus width for a given space.<\/p>\n
* Fixed incorrect address bus widths on the I8039 CPU core.<\/p>\n
* Added a TMS32026 entry and reworked the handling of internal RAM to work
\nin the new system. This means the Taito Air system is now back to its former
\nstate.<\/p>\n
* Fixed the memory handlers in gottlieb.c to use address space masking
\ninstead of mapping extra mirrored areas.<\/p>\n
General Source Changes
\n———————-<\/p>\n
Some ST-V alpha blending [Mariusz Wojcieszek]
\n\tsee bakubaku playfield, mausuke playfield, shienryu clouds etc.<\/p>\n
Huge Namco Mappy Era Systems Update [Nicola Salmoria]<\/p>\n
– Most games merged into mappy.c, so the following files are no longer
\nneeded:
\ndrivers\/superpac.c vidhrdw\/superpac.c machine\/superpac.c
\ndrivers\/grobda.c vidhrdw\/grobda.c machine\/grobda.c
\ndrivers\/phozon.c vidhrdw\/phozon.c machine\/phozon.c
\nmachine\/mappy.c
\n– one new file: machine\/namcoio.c. This implements the 56XX and 58XX I\/O
\ncustom chips used by all the games. The I\/O chips are now accurately
\nsimulated, fixing subtle nuances during normal operation and not-so-subtle
\nones in test mode; e.g. you can now scroll the grid in Mappy’s test mode.
\n– fixed and verified inputs and dip switches for all games.
\n– found and documented the Easter eggs in all games except Toy Pop.
\n– fixed interrupt and sub cpu reset handling, which caused some games to
\nreport errors when entering test mode without resetting.
\n– added detailed memory maps where schematics allowed.
\n– all drivers converted to tilemaps
\n– cocktail mode in Phozon, and non-hack implementation in Gaplus
\n– fixed priorities in superpac. With the previos code, ghost eyes had
\npriority over everything else, even other sprites, giving wrong results when
\ntwo ghosts overlapped.
\n– round advance dip switch in Gaplus available & working
\n– fixed a missing sound effect in Phozon that nobody would have ever
\nnoticed.
\n– MAME testers bugs fixed:
\nmappy37b4gre
\nmappy37b8gre
\nphozon37b8gre
\nmotos054gre
\n– I’ve also included a revised warpwarp.c, converted to tilemaps and with
\nthe Warp Warp Easter egg documented.
\n-Some slight SH-2 speedups [Carlos Santillan]
\n-Improved sound in Miss Bubble 2 and Power Instinct (bootleg set 2) [Quench]
\n-Discrete Sound in Sky Diver [Derrick Renaud]
\n-Attempted to Improve Birdie Try sprites [Pierpaolo Prazzoli]
\n-Fixed last minute compgolf bug in u3 [Pierpaolo Prazzoli]
\n-Dipswitch in cross Pang [Pierpaolo Prazzoli]<\/h5>\n
A noter que 2 roms neogeo (kof98k et aof2a) ont \u00e9t\u00e9 ajout\u00e9, d’apr\u00e8s les dumps de Hpman de la team NeoArcadia ;)))<\/p>\n
Les Dats de Nebula sont \u00e9galement \u00e0 jours (kof98k ne fonctionne pas pour le moment, c’est temporaire).<\/p>\n