Aros68k viability on real Amiga hardware

[wawa]

Originally Posted by Dunny

Indeed, I've seen that - and it's nice that they are progressing. But needing a Vampire to run it? That's a stretch, but at least it's a start.

you dont need vampire to run it..
as i might have mentioned already,my current test setup is a4000/3640/16mb. i readded p4 yesterday.

aros is currently compiled for plain 68000 and without optimizations for speed. and it certainly contains a lot of opportunities for improvement.

work is being done towards it. just look at the two past weeks highlights:

09.05.2017 - 15.05.2017:
- adapt code for const TagItem changes (neil)
- network card improvements (neil)
- debug build now stores debug info in .dbg files (Kalamatee)
- InitKickMemDiag fixed (twillen)
- Rom image size check added (Kalamatee)
- amiga-m68k rom fixes, a1200 should boot again (twilen)
- m68k rom size optimization (Kalamatee)
- cd-startup improvements (Kalamatee)
- add separate gifanim datatype descriptor)

16.05.2017 - 22.05.2017:

- gifanim datatype descriptor added (Kalamatee)
- autodocs updated (mazze)
- AddDataTypes target cpu awareness added (Kalamatee)
- cia.h defines added (Kalamatee)
- lowlevel.library m68k timer support added (Kalamatee)
- KrnIsSuper m68k optimized (twilen)
- m68k workarounds for bad software (Oktagon controller) added (twillen)
- m68k rom init fixes (twilen)
- developer files for InstallAROS fixed (neil)
- new mirrors for third party code added (neil)

thats while people pointlessly quarrel on forums, and companies threat each other.

[grelbfarlk]

I've been working out the kinks with my AROS install for preparation in an A4000 060, Piccolo, AD516, Fastlane and something else that I can't remember. Any idea why 16-bit modes wouldn't be showing up in WinUAE with Piccolo emulation? Also the corrupting of volumes has me a little worried, has this happened to anyone else or is just good old crappy FFS to blame?

[wawa]

Originally Posted by grelbfarlk

Any idea why 16-bit modes wouldn't be showing up in WinUAE with Piccolo emulation?

with aos or with aros or both? i dont hav an idea actually. never owned a piccolo. for aos maybe there is some tooltype in the monitor file like ESSENTIAL that can be changed to ALL. wouldnt affect aros though.

Also the corrupting of volumes has me a little worried, has this happened to anyone else or is just good old crappy FFS to blame?

you need to be prepared for something like this while testing. however im pretty brutal with my gear and im constantly swapping disks and cfs, and i dont remember when i lost any data or partition like that.

if you are not sure about ffs, use pfs3, just stay away from sfs, as its buggy.

[wawa]

since today aros is detecting thew whole range of 68k cpu variants inclusive 080/apollo core. some appropriate asm patches are enabled already, depending on cpu model detected. further poatches have sped up gfx operation to the point that im testing it to be within an acceptable range on an a4000/040/25mhz. still work in progress..

also i see that stepatch doesnt need to be enabled in the s-s manually anymore. thats just few minutes ago.

aha, and setting up the screen resolution defaults now to sane values. there shouldnt be any crash anymore due to bogus ones.

[wawa]

more complete overview of features implemented this week:

- .dbg files are now ignored by AROS (Kalamatee)
- lowlevel.library m68k work continues (Kalamatee)
- Gfx Bitmap-Class optimization (Kalamatee)
- m68k type detection / optimizations (Kalamatee)
- fix undefined Locale's loc_MeasuringSystem (pollux)
- Intuition Screen fixes (Kalamatee)
- Apollo'080 detection (Kalamatee, twilen)
- MUI window class compatibility enhanced (neil)
- run setpatch from startup-sequence (Kalamatee)

[ShK]

A quick Apollo'080 detection test: [ Show youtube player ]

[wawa]

cool. you booted with no problem? what model do you have? 600? i cannot judge it by the video. michael seems to have boot issues with his a500 version.

looks pretty fast even with decoration. kalamatee provided a number of speedups few days ago. might even be usable on aga, however then id disable decoration.

ah, i see, its a 500er version. wonder whats michaels problem then.

[Marlon_]

Originally Posted by wawa

cool. you booted with no problem? what model do you have? 600? i cannot judge it by the video. michale seesm to have boot issues with his a500 version.

Video says A2000 in the title.

[ShK]

Originally Posted by wawa

cool. you booted with no problem? what model do you have? 600? i cannot judge it by the video. michael seems to have boot issues with his a500 version.

looks pretty fast even with decoration. kalamatee provided a number of speedups few days ago. might even be usable on aga, however then id disable decoration.

ah, i see, its a 500er version. wonder whats michaels problem then.

V500 in an A2000 + ZorroII PicassoIV (1024x768/16). Yes, I think that IDE worked better a few build ago on bootup. It seems again to get stuck time to time on IDE drive recognition.

[wawa]

too excited to look right. i thought it may be apollo hdmi too. you can use p96 driver as you certainly know.

what concerns ide, i dont think it worked better few days ago, because i dont see changes related. it is flaky. i have a problem with a4000, but its completely fine on a1200. we are working on it whole day already. ;/

[wawa]

@shk
aha it was michael who built the rom for you. i stand corrected.
there seem to be timing problems with ata.device.

[grond]

Originally Posted by wawa

i thought it may be apollo hdmi too. you can use p96 driver as you certainly know.

It looked quite okay indeed. Now we only need that accelerated Vampire RTG driver...

[wawa]

Originally Posted by grond

It looked quite okay indeed. Now we only need that accelerated Vampire RTG driver...

we still need to iron out some bugs. among others have aros boot on each and every amiga, which was annoying issue, but there are few others rather obvious ones. also we probably need to "improve" looks, or at least have lighter and more amiga like theme, for those who damand it.

a video of aros running both its own native as well as amiga apps at the same time might be more convincing.

[grond]

Originally Posted by wawa

we still need to iron out some bugs. among others have aros boot on each and every amiga, which was annoying issue, but there are few others rather obvious ones.

I think that once these bugs have been fixed and there is an accelerated V2 graphics driver, we will have some Vampire AROS distribution with all the Vampire specific software preinstalled. Of course, anyone would be free to replace this with whatever pleases him but it would be nice to have a common denominator whenever somebody has problems with some part of the software.

[Locutus]

For 68k i think disabling theming in the stockbuild totally would be easiest.

I wrote a shell script that removes all of those items from the daily build iso image and constructs a disk image.

This is however the wrong place to do such a thing, a aros-lowend build target is what should in all likelihood exist.

[wawa]

Originally Posted by Locutus

For 68k i think disabling theming in the stockbuild totally would be easiest.

that can be done removing an env variable or un-checking an option in appearance prefs. however it wont give you different icon set, only the themeing will be off. replacing icon theme is easiest done at compile time, i guess. however i dont have anything ready just yet to try it out properly.

im fighting with amiga icon editors, realizing that there is none completely up to task. what a shame.

[Dunny]

I still think that the baseline for AROS compatibility in 68k would be to reproduce the functionality and looks of a vanilla 3.1 install running on an unexpanded A1200, with the same performance of a real unexpanded A1200. Anything else on top of that which might degrade performance such as higher colour depths, fancier icons, MUI/Zune etc would be at the user's discretion rather than built-in.

I'm sure it's achievable, but it's unfortunate that my 68k coding skills are pretty much non-existent these days. I am seriously considering getting involved in the x86 side of things, but that would mean that I need hardware that can run AROS - and although I have 6 PCs here, AROS won't run on any of them.

[wawa]

Originally Posted by Dunny

I'm sure it's achievable, but it's unfortunate that my 68k coding skills are pretty much non-existent these days. I am seriously considering getting involved in the x86 side of things, but that would mean that I need hardware that can run AROS - and although I have 6 PCs here, AROS won't run on any of them.

pretty every aros developer works primarly with a hosted version for development and testing. except you want develop particular device drivers perhaps. similarly i am testing with uae first and then with actual amigas.

[wawa]

Originally Posted by Dunny

I still think that the baseline for AROS compatibility in 68k would be to reproduce the functionality and looks of a vanilla 3.1 install running on an unexpanded A1200, with the same performance of a real unexpanded A1200.

this is exactly what doesnt make sense to me. staying compatible to that functionality, yes, this is our goal, restricting ourselves to it, isnt. if you want vanilla 3.1 functionality use the kickstart and the os your amiga has been delivered with. it is covered.

[Dunny]

Originally Posted by wawa

this is exactly what doesnt make sense to me. staying compatible to that functionality, yes, this is our goal, restricting ourselves to it, isnt. if you want vanilla 3.1 functionality use the kickstart and the os your amiga has been delivered with. it is covered.

I didn't say we should be restricted to A1200 unexpanded compatibility - just that we should be able to achieve that first. Anything else should be added at the user's discretion.

Having read your previous posts here, your main goal in your Amiga life at present is a fully open-sourced AmigaOS - a noble goal, and laudable. If that can't be achieved using the original software from Commode, then AROS is the logical next step. There really should be an AROS install that comes with a ROM image and five or six floppies, that replicates the original 3.1 KS/WB functionality.

However, if the default install requires a minimum of an accelerated and expanded Amiga, then it's not a replacement for quite a lot of users.