The commit message for 515ae3f081 (was SVN
r14679 at the time) doesn't provide a justification for changing the
periodic interrupt frequencies, implying it's just cleaning up use of
deprecated functionality. The change from 120 Hz to 240 Hz for systems
using the "lhb" configuration caused dbc, lhb33c and ryukobou to stop
recognising coins reliably (MT05678). Despite the comments added to the
machine configuration functions, changing the frequency of these timers
does not change the music tempo.
This changes the periodic timer frequencies back to what they were
previously. The timing is still suspect. We really need someone to
hook some of these games up in a logic analyser to get proper video
timings and work out where the periodic interrupts actually come from.
Sadly, portaudio is limited to the worst case of all OSes audio
supports (and ALSA est rather bad) and does not yet handle hotplug.
So the result is quite limited.
Also, the devices when running portaudio over emulated alsa over
pulseaudio are quite ugly. Still works though.
------------------------------
AV-Mahjong Yanchana Koneko (Japan Ver 1.01 1990/08/23) [Hammy]
Paint 'N Puzzle Super [Tylar R]
Piaoliang Jinhua [little0]
Piaoliang Jinhua 2 [little0]
New clones marked not working
-----------------------------
Shuang Long Qiang Zhu 2 VS (China, VS203J, set2) [little0]
- hash/a7800.xml: more PCB and ROM label updates [Guru]
- sega/segae.cpp, taito/taito_z.cpp: documentation updates [Brian Troha]
Just log a message if a game tries to blit to an out-of-range layer for
the current configuration.
Also set vertical blanking time and total lines based on measurements
for nkishusp, and always blit to the whole 512*256 area as in principle
the CPU can read areas that aren't visible on-screen.
------------------
Block Fever [Dilweed, smf]
Sound frequencies are estimated from a video of the real game, as schematics do not exist.
The clickable artwork colors were taken from the same video.
An output is pulsed when you finish a game after scoring enough points to replay, it's purpose is unknown.
Better pictures of the PCB are required to confirm the layout.
Fixes numerous graphical glitches, including missing characters in
service mode (MT08993), bad colours on vbowl game over screen, glitchy
"ODDS" banner remaining on-screen in xymg, incorrect colours on
copyright years on titles screens for lhb and clones, and more.
There's only enough video RAM for four 8-bit layers. The "depth"
register configures how many 8-bit layers are required (zero to four),
and the remaining buffers are treated as pairs of 4-bit layers. For
example xymg uses one 8-bit layer and six 4-bit layers.
The CPU always sees an 8-bit view of video memory, and must handle
updating the correct nybble to draw to an individual 4-bit layer. The
blitter always uses 4-bit source data for 4-bit layers, but can use
4-bit or 8-bit source data for 8-bit layers (it also supports 5-bit
source mode with the top bit read from a separate ROM area).
New working clones
------------------
G-LOC Air Battle (Japan) [Team Japump!!!]
Power Drift (Japan, Rev C) [Team Japump!!!]
- changed old 'pdriftj' to 'pdriftjb'
New NOT WORKING machines
------------------------
Disney Princess Dress Mania TV (Japan) [David Haywood, Team Europe]
XaviX Bowling (XaviXPORT, PT2-BWL-11, Japan) [David Haywood, Team Europe]
For the wininfos that participate as individual panes in the main console window, their viewinfos are referenced via the m_views[] array with hard-coded indices of 0, 1, and 2.
- Define & use enums for those indices (VIEW_IDX_DISASM, VIEW_IDX_STATE, VIEW_IDX_CONSOLE)
- Change MAX_VIEWS from a hard-coded constant of 4 to the last enum in that list, which changes it to 3. (I don't think it should have been 4 in the first place.)
- wininfos that NEVER appear as panes in the main console window (e.g., logwin_info, pointswin_info, etc.) continue to use m_views[0]
