agranlund wrote: Thu May 07, 2020 2:40 pm I think that was it really, pretty much just IDE and the RAM location
As far as I know, you should be able to keep going north to cover all the memory.
At least EmuTos starts reading at 0x01000000 and just keeps going until it hits a bus error in order to figure out how much of it there is.
(A quick glance at the Atari TOS sources suggests it does the same but it's all uncommented asm so I wouldn't bet my life on it)
Ok.
I think this covers the basic 64Mb version ...
Code: Select all
assign ram_decode = ({A[31:24]} != {8'h01}) & ({A[31:25]} != {7'd1}) & ({A[31:26]} != {6'd1});
Might only give 48Mb... good for a start.
I think thats covered.agranlund wrote: Thu May 07, 2020 2:40 pm I did something in bus_delay.v to get it playing nice with the Blitter but I'm guessing this is irrelevant in the new design anyway.
Now this is kinda interesting because the TF536 can do IRQ injection and patch up accesses to anything that reports them. It'd be interesting to fix this.agranlund wrote: Thu May 07, 2020 2:40 pm Oh, and Atari TOS requires the IDE IRQ in order to work, we usually solder it to pin10 of the ASCI port as it's very easy to get to.
I exposed that signal to the INT2 pad on the TF534 and ifdef'd out the Amiga INT2 related things.
But EmuTOS doesn't seem to care and I think that one will work regardless.
I don't really use Atari TOS much anymore, EmuTOS is so much better and requires less "convincing" to work with modified machines