Another PiStorm Blog

[stephen_usher]

This is a better set-up. Unfortunately no room to have a dedicated monitor for the ST so having to switch the TV inputs.

IMG_3711.jpeg

I found that the mask for the status needed to be changed as when the emulator was in 68030 mode the emulation wouldn't go into reset when the reset button was pressed as it was using a few more status bits, so I changed it from 0x00FF00 to 0x0FF000.

I think part of the issue of loading Frontbench from the UltraSATAN is that the DMA isn't working reliably enough. Trying to load the program into ALT-RAM leads to bit errors in the code -> illegal instruction or bus error. It works if loaded into ST RAM, so no multiple copying and transfer over the PS->Pi link, which doesn't seem at all reliable.

[Steve]

Glad to see you are making some incremental progress :)

[stephen_usher]

Of course, the fact that there's corruption at all means that it's not right. This work around is just masking a real problem.

[stephen_usher]

Increasing the "loopcount" variable to the max (1200 I've set it to) I can now get Frontbench to run under the 68020 emulation with ALT-RAM using EmuTOS.

Getting ~3000 frames per run. (Approximately the same as a 50MHz TF536 then.)

IMG_3712.jpeg

Note: If I go higher than 1200 loopcount things start falling over again.

[stephen_usher]

OK, after running for a while the machine's become unstable. It seems that disk I/O is now unreliable, but not if I switch over to the 68000.

I may see if I can get Frontbench onto a floppy and then remove the UltraSATAN from the equation as I've had problems with that in the past.

[Steve]

Yeah I definitely would not be bringing ACSI / DMA into the equation right now. Why not just make yourself a disk image in Hatari with all the tools you need, then mount it inside the emulator as a virtual IDE drive.

[stephen_usher]

Well, with the latest unreleased firmware and emulator the machine is a lot more stable but it's the ACSI which is the main issue.

I'm booting off floppy and I can then set loopcycle down to 1 safely. Performance is hit (Frontbench scores drop by about a third) but it's working.

I've now switched to using the on-board ROM and that's working fine as well. Surprisingly there seems to be a Frontbench speed-up when using the real ROM rather than the emulated one.

[exxos]

Well, with the latest unreleased firmware and emulator the machine is a lot more stable but it's the ACSI which is the main issue.

Did not know there was unreleased firmware. What's the difference between that and the previous release one which likely was downloaded in my setup ?

[stephen_usher]

I'm not sure. The ataritest program has the clock option and that's all the detail I know as I've only been given binaries (on the PiStorm Discord).

[stephen_usher]

There's a new Github branch of the code now: https://github.com/gotaproblem/pistorm- ... ee/may2024