Ja, danke, ist risc8.mp3

Good news, the high pitched beeps in Tasc R30 and RISC 2500/Mephisto Montreux is fixed.

I also found another issue I'd like to mention:
King 2.50(R30 version) is protected against RAM upgrades. If you install more than 512KB RAM, it will still detect the RAM expansion. But it limits RAM usage to only 128KB. So, if a person wanted to make their R30 stronger by installing more RAM, King 2.50 will actually become weaker.

King 2.2x will happily eat all hash RAM expansion you give it though.

And it appears Gideon never uses the hash RAM above 128KB. I can only speculate, maybe the prototypes of Tasc R30 only had 128KB RAM.

The above finding also means, anyone that upgraded their Tasc R40 with the R30 version of King 2.50, will only be able to use 128KB RAM.

Speaking of Tasc R40, is the speaker pitch the same as R30? Or did they increase the CPU clock divider?

https://tsk-tsk.net/net/temp/riscs/tasc30mhz.mp3 (sounds exactly like real R30)
https://tsk-tsk.net/net/temp/riscs/tasc40mhz.mp3 (higher pitch)

That's interesting (and sound a bit strange).
How did you find this? By disassembling the code (using the debugger) or by looking at the RAM usage?

Could this problem be solved by patching the 2.50 ROM, so that it uses at least the usual 512kB when installing more RAM?

Yes, I was viewing the RAM in the debugger for another reason and noticed it.
I suppose they added a few checks here and there to artificially limit the RAM. And also a ROM checksum that we already know of. It can be patched if someone has experience with ARM assembly code.

Hat jemand ein ton recording von Tasc R40, oder viellecht ein Youtube video?

Will be quite hard to get such a recording, because only very few people have this machine.

One other question:
I've now ported your new Risc2500 and TascR30 driver to my older MAME version, and there's one difference in the display for the Risc2500:
although I'm exactly using your new driver and layout in version 0.220, the upper part of the LCD looks quite different in this version than in 0.230 - I've attached the 2 screens.

There must have changed something in MAME between 0.220 and 0.230, that causes this difference - do you know which files may be responsible for this?
(the left picture is from 0.220, the right one from 0.225 and 0.230)

Miniaturansicht angehängter Grafiken

   

For me, it looks like the left one. But I can make it look like the one on the right with for example this:
mame risc2500 -video gdi
mame risc2500 -video d3d -nofilter
mame risc2500 -video opengl -prescale 6

Double check if there's anything in the configuration file like that?

Aah, that makes the difference - many thanks!
It's the option 'filter' in risc2500.ini - in 0.220 it was set to 1 (left picture) and in 0.230 the default value was 0 (right picture).

One more info about the Saitek RISC2500:
with the new driver the boot process takes much longer than with the previous versions!
When setting the RAM size to 2MB, the old version needed ~4 sec to boot, but the new version needs ~10 sec.
I don't know if this is normal!?

Or it means it was too fast before. The RAM test routine is in ROM, and the chesscomputer cpu slows down when it's in the ROM region.