Which 68060.library?

[nicholas]

most what i ever needed thors 060 and mmu libraries was his muforce. now, aros (68k) has already both 060 library functionality (within the 680x0.library, based on the same open sources all other 060 libs are based on) and enforcer functionality to be turned on/off at boot time. that said having thor as consultant on above matters (and more) would certainly be quite a gain for aros team. but i doubt he has enough time and interest, especially his doubt afair was always if aros can become compatible enough.

With PeterK's icon.library, Matt Hey's CopyMem routines, ThoR's stuff, Mathias Henze's HSMathLibs and Oliver Robert's WarpDT datatypes, AROS m68k would be much better than it is now.

It's a shame the dos.library/scsi.device/graphics.library stuff on EAB can't be used as they are Resourced from the copyrighted CBM versions.

[wawrzon]

With PeterK's icon.library, Matt Hey's CopyMem routines, ThoR's stuff, Mathias Henze's HSMathLibs and Oliver Robert's WarpDT datatypes, AROS m68k would be much better than it is now.

It's a shame the dos.library/scsi.device/graphics.library stuff on EAB can't be used as they are Resourced from the copyrighted CBM versions.

im not sure if peter.k icon lib wasnt resourced as well, only that the complete code has been replaced. now i dont know what the situation legally is. however as the work was not based on any sources, only on reverse engineering of binary and since the current library does not contain copyrighted code its okay i guess.

its possible to use this library anyway and it gives nice speedup.

what concerns math libs i have never been sure how much do they improve upon genuine ones. at least integer wise aros is currently on the same level as aos afair. have to check for fp.

as for warp datatypes these are commercial, you can surely use them under aros 68k as you do on aos, no doubt, you just need to buy them.

all in all there are few aspects aros has serious performance problem. one of them has just been traced doen to a particular function within graphics library, seems toni has problems to improve it though. i will try to make it more public on the dev list.

[nicholas]

im not sure if peter.k icon lib wasnt resourced as well, only that the complete code has been replaced. now i dont know what the situation legally is. however as the work was not based on any sources, only on reverse engineering of binary and since the current library does not contain copyrighted code its okay i guess.

its possible to use this library anyway and it gives nice speedup.

what concerns math libs i have never been sure how much do they improve upon genuine ones. at least integer wise aros is currently on the same level as aos afair. have to check for fp.

as for warp datatypes these are commercial, you can surely use them under aros 68k as you do on aos, no doubt, you just need to buy them.

all in all there are few aspects aros has serious performance problem. one of them has just been traced doen to a particular function within graphics library, seems toni has problems to improve it though. i will try to make it more public on the dev list.

Perhaps Ratte & Cosmos might be abe to help with the graphics.library problem?

[guest11527]

@ThoR

hi! have you ever planned opensourcing your MMU.library & tools? I guess it could be a great addition to AROS and would probably bring some new life to our classics :-) Maybe somebody like Matthey could add some improvements, who knows.

I would prefer a different model here as I see a lot of needless branches and "headless" discussion going on in the open source world. That does not mean that I don't want to provide sources, but it means that I would prefer someone to feel responsible and "wear the hat" as we say here.

That said, if you believe you can and want to contribute to the project, want to make some changes that are well motivated, please let me know what the plans are, and allow me to review changes before publishing any future revision, and I'll send you sources. This is the least problem.

So for example, just for the purpose of "making things faster", I would prefer to see a noticable speed improvement being measured, plus an analysis that the changes made are "correct". That's all not so easy, I afraid (and please include me in not being able to write bug-free code), and exactly the reason why I'm staying away from such modifications at this time:  

a) I don't believe it makes much a difference (because all the additional glue logic to decode the instructions and get operands is dominating, not the actual math) and  

b) it's hard enough to get things right in first place - as you actually see.

I hope you understand.

[guest11527]

@thomas: pls. update your lha-archiver, your lib has a 1980-datestamp

You just found another defect in my defect-enough A2000... )-: Real-time clock battery is dead. (Despite the already known defects of the disk drives being dead, the CDRW being dead and the omniscsi/GVP scsi card being defunct).

[nicholas]

You just found another defect in my defect-enough A2000... )-: Real-time clock battery is dead. (Despite the already known defects of the disk drives being dead, the CDRW being dead and the omniscsi/GVP scsi card being defunct).

Time to clone the drive to a WinUAE setup now that it has 68060 and MMU emulation?

[wawrzon]

Time to clone the drive to a WinUAE setup now that it has 68060 and MMU emulation?

it is (up to?) 040 mmu emulation.

[nicholas]

it is (up to?) 040 mmu emulation.

12 January 2013, 16:33
Toni Wilen

68060 MMU emulation added, mostly same as 68040, different stack frames and 68060 restarts all bus faults, 68040 restarts only read faults (excluding MOVEM).

http://eab.abime.net/861457-post2.html

[wawrzon]

I would prefer a different model here as I see a lot of needless branches and "headless" discussion going on in the open source world. That does not mean that I don't want to provide sources, but it means that I would prefer someone to feel responsible and "wear the hat" as we say here.

i know your opinion on this, but im sure if you actually was interested, you would find a quite reasonable team behind aros. their opinions may vary but usually they are following some con-sens very close to that what a regular amiga user like me actually recognizes as "improved amiga". the platform branches are a handicap on one hand because there is no focus on particular platform but a gain on the other hand, because of cross testing and remaining future proof in a fast evolving it world. the problem is of course there is no lead, so everyone just works on what he is just interested in, and there is no way to impose anything on anyone, but this is effectively the same how it is on each and every amiga related project today (mos or os4), even if it isnt openly admitted. all in all if you handed over your sources to jason or toni im quite sure they would be treated properly, but its your choice and i don want to talk you into anything.

[ChaosLord]

Which MMU is best from the "using it under WinUAE" perspective?

[nicholas]

Which MMU is best from the "using it under WinUAE" perspective?

Toni Wilen's probably the best person to ask mate.  I don't think he has an account on here though, just on EAB.

http://eab.abime.net/members/toni-wilen-257.html

[guest11527]

Time to clone the drive to a WinUAE setup now that it has 68060 and MMU emulation?

First of all, if it runs on Windows, I don't care. I don't have windows around, and it's unlikely I will get one just for the matter of emulation. There is uae on Linux, which "sort-of" works (but not quite right).  

But all that aside, things like the 68060.library cannot be tested under emulation since an emulation has no point to *not* implement instructions - as the 68060 does. The best you probably do is pick up some bugs in the emulator as the trouble here is more to work with the low-level features of the CPU.

Greetings, Thomas

[guest11527]

i know your opinion on this, but im sure if you actually was interested, you would find a quite reasonable team behind aros. their opinions may vary but usually they are following some con-sens very close to that what a regular amiga user like me actually recognizes as "improved amiga". the platform branches are a handicap on one hand because there is no focus on particular platform but a gain on the other hand, because of cross testing and remaining future proof in a fast evolving it world. the problem is of course there is no lead, so everyone just works on what he is just interested in, and there is no way to impose anything on anyone, but this is effectively the same how it is on each and every amiga related project today (mos or os4), even if it isnt openly admitted. all in all if you handed over your sources to jason or toni im quite sure they would be treated properly, but its your choice and i don want to talk you into anything.

Well, let me ask what the benefit of this would be. For emulation, the 68060.library is really pointless. For real hardware, it is (still) in supported mode, though probably with longer turn-around times, so I have no problem with it at all when people use it with AROS - or ship it with part of AROS. This is perfectly fine. The only thing I'm asking for is if you sell(!) a CD, please send me a copy, which is - I hope - not asking for too much. (At least, I have now two Amiga Forever DVDs. (-;)

Arguably, it might be anoying that it took so long to actually look into this bug and fix it. In the old days, I had turn-around times of probably one day, now it's more one month. But it also took years to actually discover it... Feature requests and bug reports are still read, though it takes time to implement and verify. Given that I hear nothing, the interest must be really low, or things must really work well. I don't know which of the two is true,  probably more the former than the latter.  

So I guess, it really boils down to one question, and that is: What would be better under a different model, and what would it enable?

[nicholas]

First of all, if it runs on Windows, I don't care. I don't have windows around

Glad to hear it!

and it's unlikely I will get one just for the matter of emulation. There is uae on Linux, which "sort-of" works (but not quite right).

The latest versions of FS-UAE on Linux work very well now. Frode has backported lots (most?) of the WinUAE code to it and it's stable and featureful enough for me to clone an exact replica of my A3000 to install, patch and configure the OS before transferring it to the real hardware anyway.

If you haven't tried it yet I really reccomend it, it's a great piece of software.

But all that aside, things like the 68060.library cannot be tested under emulation since an emulation has no point to *not* implement instructions - as the 68060 does. The best you probably do is pick up some bugs in the emulator as the trouble here is more to work with the low-level features of the CPU.

Greetings, Thomas

Understood completely.  Thanks for all the work you've done on the Mu tools and libraries by the way, I really appreciate it and I'm always reccomending your work to other people to use. :pint:

Cheers!
Nik

[Ratte]

@THOR: Are you able to create a special 060-LIB for 68LC060-CPUs.
Eyetech released a Winner(Apollo)1260 @ 75 MHz with a special "LC"-Library.
OK, it seems to be useless for "MMU"-Libs, but it could be nice to have a alternative choice.
Sideeffect, on A1k.org some guys are working on 040/060-adapters with higher clockrates.
(Warpengine 4040 -> 4060@80MHz .. thinking about MC68060FE133 and quad-clock setting)

[wawrzon]

Well, let me ask what the benefit of this would be. For emulation, the 68060.library is really pointless. For real hardware, it is (still) in supported mode, though probably with longer turn-around times, so I have no problem with it at all when people use it with AROS - or ship it with part of AROS. This is perfectly fine. The only thing I'm asking for is if you sell(!) a CD, please send me a copy, which is - I hope - not asking for too much. (At least, I have now two Amiga Forever DVDs. (-;)

Arguably, it might be anoying that it took so long to actually look into this bug and fix it. In the old days, I had turn-around times of probably one day, now it's more one month. But it also took years to actually discover it... Feature requests and bug reports are still read, though it takes time to implement and verify. Given that I hear nothing, the interest must be really low, or things must really work well. I don't know which of the two is true,  probably more the former than the latter.  

So I guess, it really boils down to one question, and that is: What would be better under a different model, and what would it enable?

what concerns different model, i see aros as a sort of code preservation repository, where all the good parts can be stored and eventually maintained beyond the point, where they would have been discontinued by their creators otherwise. that doesnt mean that anyone can be forced to contribute, but i dont believe anything worthwhile can be done on widest understood amiga platform today by a single person, let alone anything commercial can be pulled. sure, im certain you can maintain what you have developed for another few years. i dont ask you to contribute your code to aros. aros has already 680x0.library that includes 68060.library implementation. all i could imagine to ask to review the code if need be and consult the developers not to let them run into the bugs you are aware of.