dirkwhoffmann
Memory error 68020 on launch Eye of the Beholder from HD
Hi,
I'm using MC68020 + full memory options with the new Hyperion Kickstart and WHDLoad custom HD (made by me).
With mc68010 CPU vAmiga run correctly "Eye of the Beholder", with MC68020 It hangs as below:
With 68010 on, it works (same memory and chips configuration):
vAmiga does not emulate the 68020 accurately, especially w.r.t. timing. As a result, some titles do not work.
I tested my 68020 implementation with test cases generated with Toni Wilen's cputester. Maybe some better tests could be generated with a newer version of cputester, but I am not sure if such a version exists. Right now, I don't have access to Windows, thus I cannot create test cases myself (cputester is Win-only).
Ok. Mc68010 is good enought for WHDLoad (quit option). So better to improve mc68010 emulation then. Thanks
68010 compatibility should already be really good. I upgraded one of my A500s a while ago and I can therefore run all vAmigaTS tests on a real machine.
I still remember the time around 1989 or so ... having the value of 35 euros in my pocket and wanting to buy the MC68010 to upgrade my A1000 and play around with its new instructions...
Exception 108 looks weird. I can reproduce it, but address is at a different point though also in some interrupt related function in (the emulated KS1.3) exec.library. So likely related to interrupt handling. Maybe won't fix it, but probably there's something to look at.
If you have PGP public key, I can attach an encrypted HDF here or I can email it if you want. It's a little over 5MB.
If you have PGP public key, I can attach an encrypted HDF here or I can email it if you want. It's a little over 5MB.
Cool. Here is my public key: Dirk Hoffmann (3E525EB5) – Öffentlich.asc.zip
a600_k32.hdf.gpg.zip (Note: not actually a zip file, but gpg isn't an allowed extension).
KS3.2 ROM required (and also seems like chipram needs to be limited to 1M)
Thanks! I'm able to reproduce the issue with your HDF (68020, 1 MB Chip, 512 KB Slow, 8 MB Fast):
I’m closing this issue for now, as I don’t plan to work on 68020 compatibility in the near term. A game changer would be an improved version of cputester to create better test cases (some are already available in the vAmigaTS repo, which I used heavily while developing 68020 support). Unfortunately, I couldn’t get the latest version of cputester to run on my machine.