Overview Features Coding ApolloOS Performance Forum Downloads Products Order Contact

Welcome to the Apollo Forum

This forum is for people interested in the APOLLO CPU.
Please read the forum usage manual.
Please visit our Apollo-Discord Server for support.



All TopicsNewsPerformanceGamesDemosApolloVampireAROSWorkbenchATARIReleases
Information about the Apollo CPU and FPU.

Apollo Vamp V4 Mmu Bounty - How To?page  1 2 3 4 

Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
23 Sep 2019 11:48


I agree that there are far better stuff then m68k mmu - but there are also reasons for it, as discussed. And multiple ways to skin a cat
     
      I don't insist Apollo team should do it, if other skilled coders can do it in vhl or software.
     
      If Apollo does it, when they can.
     
      Why wouldn't "mmu caring" people like me support this bounty, and I am sure you can find more ideas that improve both back compatibility or ensure future for vamps/Apollo core.
   
    You would need the same "4 points" thing, not the idea and surely, a sucesfull bounty and development and you can have your piece of pie too.
 
 
Gunnar von Boehn wrote:

    Patching the old AMIGA tool using the new MMU is magnitudes less work than changing the MMU.
 

 
  This is fully acceptable - form of MMU.lib, emulation layer or patching few selected MMU enabled AOS 3.x apps would be enough.
 
  Possibility for AROS to use new Apollo MMU too.
 
  Linux kernel that would fully use 68080-AMMX and new MMU would be best, v2 and v4 preferably. Kernel pack could also include drivers for SAGA, Vamp USB and LAN.


Gunnar von Boehn
(Apollo Team Member)
Posts 6197
24 Sep 2019 07:43


Vojin Vidanovic wrote:

Linux kernel that would fully use 68080-AMMX and new MMU would be best, v2 and v4 preferably. Kernel pack could also include drivers for SAGA, Vamp USB and LAN.

Of course one can already today run Linux on Apollo-68080.
And with 040 style MMU emulation - normal Linux should run out-of the box on Apollo-68080.

But what I'm afraid here a little bit is to "wake" wrong expectations in users.

The VAMP will run Linux faster than any other 68K can do - and there are many uses for such a Linux system. You could easily run a MySQL database on it, or an APACHE webserver. And you could even serve such a webpage of forum like we have here on a Vamp.

I do not think that OpenOffice or Firefox will be much useable on this - and it would be bad if people get wrong expectations.



J.M. Lapilainen

Posts 24
24 Sep 2019 12:50


Gunnar von Boehn wrote:

  I do not think that OpenOffice or Firefox will be much useable on this - and it would be bad if people get wrong expectations.

  Yeah, this is exactly why I've tried to be vocal about not expecting too much. People who donate under false expectations would most likely make a scene and start pissing on everyone's morning cereal when they notice they're not getting what they were expecting.
 


Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
25 Sep 2019 06:26


Gunnar von Boehn wrote:

  Of course one can already today run Linux on Apollo-68080.
  I do not think that OpenOffice or Firefox will be much useable on this - and it would be bad if people get wrong expectations.

I know there Is ucLinux but kernel Is quite pod, even distro Has Been updated 2016.

I have no illusions modem Linux will crawl. But if one starta with e.g. Ff4.0+ and abiword and hw geta faster, with proper kernel, who knows.
I AM sure that comeback of m68k Linux on vamp would Make more backports then Aros and amigaos/freemint old schoolers, that I do care for to.

So, dont avoid the penguin, please



Carlos Milán

Posts 95
28 Sep 2019 22:03


If we are talking about Linux, I think it makes more sense the other way: to get the mainstream kernel to support the Apollo Core native PMMU. I remember that a well known Linux kernel dev, EXTERNAL LINK  was here interested on that, and he even got a Vampire 500; but never anymore about him regarding this topic. He needed documentation about the PMMU implementation.

By the way and going offtopic here, this week a Linux kernel driver for the ICY Zorro Card EXTERNAL LINK was successfully commited into the mainstream Linux kernel and Linus Torvalds wrote himself about that: EXTERNAL LINK 
"Christ. Will that thing _never_ die?

But the reason I'm actually replying is not to comment on the apparent death-defying Amiga hardware scene, but to point out that you should try to fix your email configuration"

No, Amiga will never die :)


Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
28 Sep 2019 22:14


I agree Linux kernel support would be great!
  Something AEON and Acube failed to do.
 
  Even if it "costs few SA send somewhere" (Debian developers, Linus instead of his QL?)
 
  A bit of AmigaOS app patching and AROS support for PMMU and all would be fine and dandy.
 
  If any of OS 3.x "MMU using" app has sources on Aminet, maybe that could replace enforcer / be patched to use new PMMU and save us the whole old m68k MMU bounty SAGA.

Or these require 3 separate bounties? :)


Mr Niding

Posts 459
29 Sep 2019 07:17


Not to be a killjoy, but have we even any compentent Linux oriented developers amongst us that are inclined and have the time to adapt a distro to Vampire?
 
kolla did remind me about Matthew Garrett that asked about MMU a while ago, but Im not sure what became of this. Wether or not he just faded away or.
 
CLICK HERE 
Personally I use Linux Mint, alongside Windows 10 on my AMD hardware, so Im positive minded towards Linux.
But it almost feels like focusing on AROS is better..? Since its grounded in the Amiga heritage.


Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
29 Sep 2019 07:31


Both Are open source and multi arch.
 
  Aros Has few Linux componentts, os3 compatibility, compiled Mos and os4 Level apps and Linux Has way more and more modern apps, while compared to Aros Is more resource hungry.
 
  Adopting Aros will bring amigans.
 
  Having Linux too servers for mainstream acceptance. As well as Linux Had a history with m68k until Motorola killed it.

Vamp COULD correct that a bit :-)
 
 


Ronnie Beck

Posts 37
29 Sep 2019 10:11


Mr Niding wrote:

But it almost feels like focusing on AROS is better..? Since its grounded in the Amiga heritage.

And amen to that.  I want to use Amiga software on an Amiga OS on my vampire.
  Enough with the whole mmu/we-must-have-linux discussion.  Please, i beg you.  Enough.


Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
29 Sep 2019 11:35


Ok lets focus on Aros first (even it Has x11)
 
  This time next year we might be penguins too, not only amiga cats. Penguin versatility Is that it Can work on toaster if it Has cpu or microcontroller


Gunnar von Boehn
(Apollo Team Member)
Posts 6197
29 Sep 2019 13:34


Vojin Vidanovic wrote:

Penguin versatility Is that it Can work on toaster i

Toaster with MMU?



Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
29 Sep 2019 17:23


As we know, there are even MMU less kernels.

But, yes, toasters should have MMU if they have RAM, since 030.


Barana V
(Needs Verification)
Posts 59/ 1
22 Oct 2019 11:25


Gunnar von Boehn wrote:

 
Vojin Vidanovic wrote:

  Penguin versatility Is that it Can work on toaster i
 

  Toaster with MMU?
 
 

  Marmalade Management Unit.
  Peanut and Marmalade management Unit. PMMU
  You don't know this Gunnar? Oh Dear! ;)


Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
22 Oct 2019 11:31


Having Linux working on a toaster was a inside joke
    on more things getting ARMs with Internet of things, plus the Linuxes multiarch ability and versatility.
 
  Could and should be anywhere, almost like Amiga Anywhere, just more carefully executed. m68ks influence was minor: one of first supported and first ditched archs, plus that Linus learned coding on an Sinclair QL. Ah, Spectrum believers, PC-Win was not always upgrade path for them!
     
      Personal Memorial Military Unit :)
   
      Well, maps chip as fast and kicks ROM all around.
     
      Past argument was that m68k MMU emulation would mean a slowdown.
     
      I hope better could be done, not a priority, that is all. Linux is a added compatibility bonus. Once bounty was mentioned "as a tool to speed things up", or similar (bump on priority list and workload)


Gunnar von Boehn
(Apollo Team Member)
Posts 6197
22 Oct 2019 14:03


Vojin Vidanovic wrote:

  bounty was mentioned "as a tool to speed things up", or similar (bump on priority list and workload)
 

 
  Lets be crystal clear.
 
1) yes a backward support layer (040like) could be done
 
2) We need define the goal
    2A) just do the minimum to get Enforcer running without software patch
    2B) Make it proper so that also Linux runs
 
3) Its possible but it will be significant work, 4 month timerange
 
4) A proper bounty will surely help people to put focus in this



Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
22 Oct 2019 14:34


Gunnar von Boehn wrote:

  3) Its possible but it will be significant work, 4 month timerange
  4) A proper bounty will surely help people to put focus in this

Goal is surely to run Linux or any AOS software that require Linux. That potentially helps Mac and Atari emulation, as they used MMU too. Kills last of m68k incompatibilities and a nagsayer argument.

It does not have to be 4 months now, but hope can be cared for.
Not more important then warp3d core. But not much less either.


Gunnar von Boehn
(Apollo Team Member)
Posts 6197
22 Oct 2019 14:56


Vojin Vidanovic wrote:

Kills last of m68k incompatibilities and a nagsayer argument.

Lets please leave this naysayer topics away.
naysayers can always find an argument.
And if you fix a CPU bug - they might say that fixing is bad as the program now not crashes anymore as it used to be.

Lets focus on the positive goal to enable more software.


Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
22 Oct 2019 15:08


Gunnar von Boehn wrote:

    Lets focus on the positive goal to enable more software.
   

   
    Full m68k fix should:
    - Eliminate some incompatibilities in AOS software (030-060 software that requires MMU, VMEM options in programs and LIBS that use MMU etc.)
    - Enables direct Atari ST/TT emulation (only Yamaha needed)
    - Enables direct Mac Classic m68k emulation (just the gfx modes)
    - Enables m68k Linux and enables 080 to be seen as m68k arch revival from Linux POV
- Aids emulation of other m68k based systems that had 030+ and MMU
   


Wawa T

Posts 695
22 Oct 2019 15:14


Gunnar von Boehn wrote:

  naysayers can always find an argument.
 

  good observation. that from someone, that has been labelled naysayer places over and over again.
 

  And if you fix a CPU bug - they might say that fixing is bad as the program now not crashes anymore as it used to be.

  or crashes on workaround now. thats a can of worms. but being a originally in a mmu supporter camp, just not as vocal, i lacked insight and interest, i wouldnt nail stuff on it.
 


Gunnar von Boehn
(Apollo Team Member)
Posts 6197
22 Oct 2019 15:38


Vojin Vidanovic wrote:

 
Gunnar von Boehn wrote:

      Lets focus on the positive goal to enable more software.
     

     
      Full m68k fix should:
      - Eliminate some incompatibilities in AOS software (030-060 software that requires MMU, VMEM options in programs and LIBS that use MMU etc.)
      - Enables direct Atari ST/TT emulation (only Yamaha needed)
      - Enables direct Mac Classic m68k emulation (just the gfx modes)
      - Enables m68k Linux and enables 080 to be seen as m68k arch revival from Linux POV
  - Aids emulation of other m68k based systems that had 030+ and MMU
     
 

 
 
Sorry but your post does not help to clarify here.
Your invisioned goals are way to fuzzy like "Enable Mac" ...
Its not MMU which is needed to this.
Also for ATARI is not the MMU what is missing for classic ATARI software.
I find your post misleading of users.

 
Also asking for VMEM is not clever.
VMEM on flash drives like we use today (CF/SDcard) will result in much faster death of the drive we use.
 
So we definitely do not want to users use VMEM and that they kill their drives by this.
 
 
 
 
 
 

posts 75page  1 2 3 4