I had previously used a windows emulator through wine (WinX68kHighSpeed_eng) and had beautiful output from the game Gemini Wing playing on my CM-64, since it has options to meld CM-32P PCM sounds with the CM-32L synth sounds.
Now, I've been trying to use USB-MIDI from the MiSTer's X68000 core and have been getting stuck notes galore in Gemini Wing with any of the CM-64 option keys F2-F4 held on game boot. This even happens when starting it in CM-32L/MT-32 mode by holding F1 while booting the game. The main synth sound freezes the moment it begins playing in the very first level, while the drums continue but otherwise only certain short tidbits are heard.
It's not an issue with MIDILink, as the USER IO port running my MiSTer MIDI 1.4 adapter acts the same. I've also tried a couple Roland branded MIDI cables, as well as multiple USB adapters without any difference (Edirol UM-1S, Edirol UA-4FX, and Yamaha UW500). Finally, I've tested it with my CM-500 in its mode B (CM-64 simulation) too but it's the same.
The X68000 core is set through the OSD to use "USB" for the MIDILink option, and 31250 BAUD (with no other BAUD rate choices shown, but that should be correct for the X68000).
On the other hand, Acumajou Dracula sounds perfect, through all of these setups and cables.
A tip if you're trying this: Sharp X68000 games boot fast from .HDF SASI hard drive images.
Next, I investigated the X68000 core's Gemini Wing MIDI output by running it through a computer running aseqdump, and I get drastically corrupted SysEx messages. During the initial loading of the F4-held SN-U110-07 mode which should have most SysEx messages as 256 bytes each (according to the emulator's output and verified CM-64 playback), from the MiSTer they are all 22 bytes each and they don't even appear to be messages for the MT-32 family, which should begin with "F0 41 10 16 12".
But that's not all I noticed: After the game has been opened from the menu by selecting "Game Start", normally as the music starts for the first level there are a few SysEx messages in certain parts of that level's song which repeat when the song loops. They don't appear to play at all on the MiSTer's X68000 core.
That's the easily noticed corruption for me to see, but I suspect it may even be corrupting the note off messages as that's the reason I questioned it working in the first place as that aspect is obvious though audio.
I had previously used a windows emulator through wine (WinX68kHighSpeed_eng) and had beautiful output from the game Gemini Wing playing on my CM-64, since it has options to meld CM-32P PCM sounds with the CM-32L synth sounds.
Now, I've been trying to use USB-MIDI from the MiSTer's X68000 core and have been getting stuck notes galore in Gemini Wing with any of the CM-64 option keys F2-F4 held on game boot. This even happens when starting it in CM-32L/MT-32 mode by holding F1 while booting the game. The main synth sound freezes the moment it begins playing in the very first level, while the drums continue but otherwise only certain short tidbits are heard.
It's not an issue with MIDILink, as the USER IO port running my MiSTer MIDI 1.4 adapter acts the same. I've also tried a couple Roland branded MIDI cables, as well as multiple USB adapters without any difference (Edirol UM-1S, Edirol UA-4FX, and Yamaha UW500). Finally, I've tested it with my CM-500 in its mode B (CM-64 simulation) too but it's the same.
The X68000 core is set through the OSD to use "USB" for the MIDILink option, and 31250 BAUD (with no other BAUD rate choices shown, but that should be correct for the X68000).
On the other hand, Acumajou Dracula sounds perfect, through all of these setups and cables.
A tip if you're trying this: Sharp X68000 games boot fast from .HDF SASI hard drive images.
Next, I investigated the X68000 core's Gemini Wing MIDI output by running it through a computer running aseqdump, and I get drastically corrupted SysEx messages. During the initial loading of the F4-held SN-U110-07 mode which should have most SysEx messages as 256 bytes each (according to the emulator's output and verified CM-64 playback), from the MiSTer they are all 22 bytes each and they don't even appear to be messages for the MT-32 family, which should begin with "F0 41 10 16 12".
A few entire-message examples of these corrupted messages follow: F0 41 10 20 00 32 06 0C 29 00 00 5A 5C 00 64 00 05 00 32 00 1B F7 F0 41 10 6E 00 32 00 0C 2E 32 00 52 61 00 00 00 00 00 32 00 0C F7 F0 41 10 61 3C 32 20 0C 49 32 64 3C 36 01 0F 64 32 3C 31 18 0C F7
An example good message (the first 256-byte Initial-SysEx message) from the emulator follows: F0 41 10 16 12 08 00 00 52 45 45 44 31 20 20 20 20 20 00 00 03 00 24 37 0B 01 01 00 21 07 00 00 00 00 00 34 00 32 32 32 32 32 00 00 00 64 0A 01 27 08 18 64 00 00 06 14 21 4D 53 2D 45 4C 00 64 4A 59 0C 19 0C 00 00 00 00 00 00 2B 64 64 64 64 24 29 0B 01 01 00 00 07 00 00 00 00 00 34 00 32 32 32 32 32 40 00 00 64 10 0E 67 06 00 08 00 00 00 00 00 00 00 00 00 00 00 5A 46 5B 0C 1B 0C 00 00 09 05 0B 0E 0F 4D 5C 62 64 24 2F 0B 01 00 00 00 07 00 00 00 00 00 34 00 32 32 32 32 32 40 17 38 64 10 05 67 06 00 00 00 00 00 00 00 00 00 00 00 00 00 46 46 5B 0C 1B 0C 00 00 09 05 0B 0E 0F 4D 5C 62 64 24 2F 0B 01 00 00 00 07 00 00 00 00 00 34 00 32 32 32 32 32 40 17 38 64 10 05 67 06 00 00 00 00 00 00 00 00 00 00 00 00 00 46 46 5B 0C 1B 0C 00 00 09 05 0B 0E 0F 4D 5C 62 64 77 F7
But that's not all I noticed: After the game has been opened from the menu by selecting "Game Start", normally as the music starts for the first level there are a few SysEx messages in certain parts of that level's song which repeat when the song loops. They don't appear to play at all on the MiSTer's X68000 core.
That's the easily noticed corruption for me to see, but I suspect it may even be corrupting the note off messages as that's the reason I questioned it working in the first place as that aspect is obvious though audio.
Thanks for any further suggestions or solutions!