Yeah, this lines up with my theory actually, which is similar to Toxie's DMA latency one above. The media access handling in Nuance is really janky. It performs the read and makes the completion callback right in the read request function, which would likely never happen on real hardware. I've had it on my TODO list to implement a background I/O thread that spoon-feeds data at roughly DVD read speeds back to the program instead, but I haven't gotten around to it. My goal is to get music streaming testing working for home brew purposes, but it's possible it could help T3K as well.I've started looking at Tempest 3000 again. Found several ways to consistently make the emulator unresponsive. I've also observed two cases of the real machine (a recently acquired Extiva) crash in the same spot the emulator will.
I believe at this point, that the issue comes down to the emulator needing to access a media file.
What I would like to ask, is there a way to somehow set a breakpoint, or log when the emulator is either not getting VSYNC commands, or when the framebuffer is no longer changing? I ask this because the emulator does not hang. It merely stops updating the framebuffer and responding to keypad. I have had a few cases where a sample would continue to play through to the end so I know audio was still being processed.


