68030/mmu Support in WinUAE
Will this ever be supported in WinUAE?
:D |
Maybe,
it's possible at least, since there is an existing MMU emu for the "Aranym" (Atari runs on any machine) already, which is based on the original UAE CPU code. So, it shouldn't be too difficult to port this piece of software back to WinUAE, but who wants to do that? And how much would you pay for ? :laughing http://aranym.sourceforge.net/ Source: http://prdownloads.sourceforge.net/aranym/ |
what would you be using the mmu support for ?
NetBSD / Amix / Linux ?? |
Debugging!!! (enforcer etc..!)
|
Playing with Shapeshifter, etc anything that uses the MMU.
I would be cool because it would exactly match my A1200. I used WinUAE to setup Workbench on my A1200 then unlha'ed it to my main partition. Are there any major apps that take advantage of an MMU on the Amiga? Dan |
030 Support is useless since 020 uses the same instructions as 030 ;)
For MMU Support look here -> http://eab.abime.net/showthread.php?t=13428 |
Yeah, this came up when I was asking for support for the early A3000's 1.4 boot ROMs (which don't currently boot due to the lack of 030 MMU). Toni's pretty much said that it's too much work for minimal benefits. It'd be nice to see at some point in the future, but I wouldn't expect to see it for a good long time, if at all.
|
Is there any instruction Difference between the 020 and the EC020?
or is it just the MMU instruction set that is missing in the EC020 ? |
I am probably repeating myself again..
68030 (or 68020 + 68851) MMUs are extremely complex compared to 68040's MMU. 68040/060 MMU emulation is possible (and already done in Aranym) but the problem is incompatibility with JIT. End result is nice but really slow toy.. 68EC020 = 24-bit address space, no MMU 68020 = 32-bit address space, no MMU 68851 is separate MMU chip for 68020 CPU (just like 68881/68882 is separate FPU chip) |
thanks :)
|
Quote:
|
Quote:
Or use ages-old Minix. But that's not a real alternative, is it. @Toni: Do not forget that there is a way to merge Dave Haynie's old 0.8.2x patch into the new source tree if someone does it. Maybe even Richard can do that, since he's very experienced with Linux UAE. Better MMU support in Linux only than nowhere. :) |
My point is: MMU in emulation is not needed. You can run Linux/Unix in PC (and you can't have HD emulation in non-AmigaOS operating systems anyway..) You don't need virtual memory in emulation. You don't need enforcer in emulation (WinUAE has build-in similar feature)
Yes, MMU emulation would be nice feature but not worth the trouble. I'll do it someday but only when I feel like doing it.. Also most likely result is only complaints about slow speed... |
Yeah, it would be more use to implement it for full A3000 compatibility, and then being able to use it for *NIX would be a nice side-effect. Even then, there's not that much utility other than being able to run A3000-specific boot ROMs. I'd like to see it, but I'm willing to wait as long as it takes, accepting the fact that it might never be implemented.
|
Quote:
|
MMU would be handy to code WHDLoad slaves properly on WinUAE too.
|
Quote:
|
for WHDLoad a MMU with much reduced function set would be sufficient.
only needed features are Pages must be invalid/readwrite/readonly, reporting of kind of access on faults and continue(like 30) or restart(40/60) instructions. |
Would it be simpler to extend the existing WinUAE debug console to have WHDLoad like functionality for memory access, "snoop" and illegal instruction trapping?
|
Quote:
access to invalid custom/cia should be no problem. access to invalid memory only partiell because uae does not know which memory is used by whdload/installed program. CBAF would not be possible |
All times are GMT +2. The time now is 19:30. |
Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.