exxos's Atari, Amiga & retro forum

alenppc wrote: 13 Mar 2019 22:06 Consoldering the CD32 was hastily built as a company saviour they probably didn’t do the usual diligence. Maybe it’s not possible to detect different revisions via software.

Indeed. I have a horrible feeling that the Rev 0 might not allow the 020 to be disabled.

What about the Akiko mirror addresses...?

Gooeyblob wrote: 13 Mar 2019 22:35 What about the Akiko mirror addresses...?

I dont follow.

Is the same Akiko identifier present at B80040, B80080...B87FC0?

Gooeyblob wrote: 13 Mar 2019 23:08 Is the same Akiko identifier present at B80040, B80080...B87FC0?

I suppose its worth checking but i would be incredibly surprised if the 64 byte mirrors had different data. Usually with these things its because they cheaped out on the address decoding.

Another address that may be worth checking between revisions are bits 0 to 22 of Config/B80024.

My gut feeling is that 020 does get disabled, but then something else in the logic is different. The behaviour is consistent, yellow screen, flashing led, black screen, flashing led, yellow screen. If both CPUs were on, I don't think we would be seeing this.

I don't know anyone else with the SX32 AND a rev 0 Akiko... but surely this would have popped up sooner if that had been the case.

alenppc wrote: 13 Mar 2019 23:39 My gut feeling is that 020 does get disabled, but then something else in the logic is different. The behaviour is consistent, yellow screen, flashing led, black screen, flashing led, yellow screen. If both CPUs were on, I don't think we would be seeing this.

I don't know anyone else with the SX32 AND a rev 0 Akiko... but surely this would have popped up sooner if that had been the case.

My thought process...

There are 3 ways the TF330 interacts with Akiko..
(i) It holds CPU_BR low for all time.
(ii) It uses EX_BR/EX_BG for bus arb in place of CPU_BR/CPU_BG.
(iii) it expects PUNT to disable all the decodes on the mobo.

The only other weirdness might be the FPU detection code in the OS. Since people are reporting DiagROM going mad when accessing Fastram i cant see how any of this relates to Akikio. Unless PUNT is non-functional somehow or the 020 is still running.

When the TF330 accesses Fastram it simply tells the Amiga to shut up.

I'm very confused at this point.

It would be helpful if there was any documentation about the differences between the rev 0 and rev A, but I can't seem to find anything.

terriblefire wrote: 13 Mar 2019 22:27
Indeed. I have a horrible feeling that the Rev 0 might not allow the 020 to be disabled.

Nope, 020 CPU is cold as women's feets, there must be something else