(I believe this to be correct, it's internal to the CPU?)
note, DERIVED_CLOCK doesn't work in cases where MCFG_CPU_MODIFY is used to modify the clock, so I've had to use MCFG_CPU_REPLACE in those cases instead.
note2, sound routing seems a bit messy, but this seems a general MAME thing, we still have an issue where if you add too many devices with their own speaker / sound additions things you get sound overflows, in this case all drivers except NES were set to a 0.50 sound level output, NES was set to 0.90, I see no clean way to do this with the sound chip inside the CPU?
n64: Simulation of cpu cycles from the skipped RDRAM initialization. This allows later SI initialization to see something closer to an expected value for the system uptime.
Missing:
- hsync is not handled as it should yet, especially missing on the zx80
- the versions with an external charrom are not handled yet
- the zx97 is weird
On the other hand, the zx81 ain't bad and the hires modes work.
Changed rsp disassembly to use MIPS ABI names for registers.
Implemented 'start valid' bit in RDP status register.
Increased scheduler quantum time to allow for better r4300 <-> RSP signaling.
PI transfers round down/truncate last bit
SI transfers signal busy while waiting for delay
Initial work to seperate VI from RDP as the actual hardware has no direct communication between them
Stop ERET instruction spamming the error log
Stop screen update from spamming the error log for using MCFG_SCREEN_VBLANK_TIME. The value set was also complete nonsense.
