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
Performance and Benchmark Results!

Apollo Hardfpu Resultspage  1 2 3 4 5 6 7 8 

Vojin Vidanovic

Posts 770
08 Oct 2017 19:58


Mallagan Bellator wrote:
 
  Why does the Amiga need Linux? There are tons of other systems that may run Linux much better. Vampire powers up the Amiga OS, AROS and Amiga software. Why transform a pair of dice for Yatzy into cards for Poker?

 
  Seeing advanced 040 setups with a lot of RAM, running Kernel 2.x style GNOME Linux, I believe Vamps could do it decently. We are not speaking of modern Linux distros, as they dont exist for m68k.
 
  Its exploit of generic m68k side of Vampire, not of its SAGA features. Boths are equally legal, as they were with APUS/Debian 3.x Amiga and Atari support, or as it is with PPC/x86 modern Linux designs that run on otherwise "Apple" and "Wintel" branded boxes. Or our phones and tablets.
 
  Just as with Amitari project, overgrow the "Amiga only" picture and see it as hardware that can be utilized in many ways.

Gunnar von Boehn wrote:

  uLinux runs certainly today on APOLLO just fine.
  But Linux is slow compared to AMIGA OS or ATARI TOS.
  And LINUX gui does today not know SAGA Screenmodes.
  So to make LINUX use truecolor on VAMP you will want to add drivers for SAGA RTG, ideally with AMMX code to make the Linux GUI faster.
  This will be some work for coders.
  While this can be done.

I would certainly try it, even as in AROS, without currently
optimized video drivers (and with hope for it). For example, Debian used to have even AGA support, which is enough for beginning
EXTERNAL LINK 
No one would try Linux for speed (compared to AmigaOS), that is not the point. Its just for few apps otherwise unavailiable, and hope a bit more power and ram v4 Vamps offer is a good step forward.

I do understand, this is not currently on roadmap. Maybe a power2people bounty one day.


Mallagan Bellator

Posts 393
08 Oct 2017 20:03


Gunnar von Boehn wrote:

Kolbjørn Barmen wrote:

  I think it is a missed opportunity,
 

 
  You spend a lot time complaining.

Lol xD PERFECT!!
If his complaining and moaning was processed by the 080, it would have over with in less than a second, despite all the "instructions" used xD

Perfect... :)


Mallagan Bellator

Posts 393
08 Oct 2017 20:30


One thing that would be cool, though, is wine (linux Windows "emulation") ported to Amiga, for the use of playing old Windows games that no longer play correctly on modern PCs


Peter Heginbotham

Posts 214
08 Oct 2017 20:34


Mallagan Bellator wrote:

One thing that would be cool, though, is wine (linux Windows "emulation") ported to Amiga, for the use of playing old Windows games that no longer play correctly on modern PCs

Why when I could use dosbox, VMware, virtualbox etc


Nixus Minimax

Posts 416
08 Oct 2017 20:40


Mallagan Bellator wrote:

One thing that would be cool, though, is wine (linux Windows "emulation") ported to Amiga

"wine" means "wine is no emulator". It simply provides software libraries to run Windows programs natively in Linux. Thus, wine requires an x86 CPU.


Mallagan Bellator

Posts 393
08 Oct 2017 21:18


Nixus Minimax wrote:

Mallagan Bellator wrote:

  One thing that would be cool, though, is wine (linux Windows "emulation") ported to Amiga
 

 
  "wine" means "wine is no emulator". It simply provides software libraries to run Windows programs natively in Linux. Thus, wine requires an x86 CPU.

AOS 4.x emulates 68k instructions natively with jit emulator. I was thinking something along those lines, as well as adding the libraries from wine. See? ^^


Kolbjørn Barmen
(Needs Verification)
Posts 219/ 2
08 Oct 2017 22:29


@Gunnar

The core aimed for V2 - does it already, or will it have the 68080 MMU?

Will it be possible for anyone with that kind of interest to use that MMU, without having to go through you first?


Kolbjørn Barmen
(Needs Verification)
Posts 219/ 2
08 Oct 2017 22:36


Mallagan Bellator wrote:

Kolbj�rn Barmen wrote:

  You are missing the point - what made Linux possible on Amiga in the first place?
 
  What was it that made it take so much longer to get Linux and the BSDs to mature on on 68k Mac compared to on Amiga, Atari etc?
 

  Interest, or lack thereof, made it take so much longer

EXTERNAL LINK


Peter Heginbotham

Posts 214
08 Oct 2017 23:31


This is a little off topic

Realistically are there going to be more than a hand full of people going to use and ancient version of Linux on an Amiga with a vampire. There are bigger coding priorities



Kolbjørn Barmen
(Needs Verification)
Posts 219/ 2
09 Oct 2017 00:03


Steve Ferrell wrote:

  If I want to run a virtual Amiga, I have WinUAE. 
 

 
You will also get it with Vampire, congrats :)

Some speculations...

There is nothing in Amiga OS that is capable of dealing with the hardware that the V4 comes with, right?

So something else is doing it, the Apollo Core firmware that is initiated on power-on, the MMU of the Apollo Core that maps the memory so that exec will find what it is looking for, where it expects to. Remember that Apollo Core is multithreaded, while Amiga OS is running on one thread, "whatever" can run on the other thread, shuffling data around in memory without anything on the "Amiga side" being involved, the MMU can be told to remap memory addresses, what was a given memory address in one moment can be pointing at something completely else the next moment. You write data to "chip RAM", and it might magically be available in fast RAM and vice versa. Awesome for speeding up operations, an I/O driver can use the other thread to do stuff behind the back of the OS.  I suppose vampire.resouce and processor.resource are provided for programmers to tap into this power.


Steve Ferrell

Posts 424
09 Oct 2017 00:16


Kolbjørn Barmen wrote:

Steve Ferrell wrote:

  If I want to run a virtual Amiga, I have WinUAE. 
 

 
  You will also get it with Vampire, congrats :)

Congrats. You're still trolling.


Vojin Vidanovic

Posts 770
09 Oct 2017 00:54


Peter Heginbotham wrote:

  This is a little off topic
  Realistically are there going to be more than a hand full of people going to use and ancient version of Linux on an Amiga with a vampire. There are bigger coding priorities
 

 
  Yes, there is even a dedicated topic for it.
 
  Same handfull might use Linux 68k, that might use EmuTOS, or Classic MacOS for various reasons, and same goes for SCUMM, DOSBox, you name it. Its better to have options then to have not.
 
  On the development of Apollo core and SAGA chipset, yes there are much much higher priorities to do, on software side for AmigaOS nothing will appear magically, thus is good to know *ALL* options.
 
  To get back on topic, I see from results Vampire v4 hard FPU can do Motorola FPU things people used to nag like this, fastest seen so far in m68k world.

Hope that will provide boost to any m68k software and that finally using 030 fpu, 040 fpu and even 060 binaries that dont hit MMU would be possible. Be it not immidiate but ironed in next few releases.

Its important because Amiga compatibility with FPU and AGA in chipset will be almost complete. CPU is quite mature, time to do the SAGA and drivers.
 


Kolbjørn Barmen
(Needs Verification)
Posts 219/ 2
09 Oct 2017 12:10


Vojin Vidanovic wrote:

We are not speaking of modern Linux distros, as they dont exist for m68k.

Why do you say so? Debian is not modern enough for you?
EXTERNAL LINK


Kolbjørn Barmen
(Needs Verification)
Posts 219/ 2
09 Oct 2017 12:34


Vojin Vidanovic wrote:

To get back on topic, I see from results Vampire v4 hard FPU can do Motorola FPU things people used to nag like this, fastest seen so far in m68k world.

Fastest claimed, yet to be seen, or are we suddenly supposed to take sysinfo seriously?

Anyways, the nag was never about FPU on v4, it was about no FPU for v2, and instead AMMX that very few have any use for. Well, now it looks like FEMU in 2.7 upwards will use AMMX, so... yay, or something.


Olaf Schoenweiss

Posts 690
09 Oct 2017 12:54


Kolla now active on another forum? Still on crusade mode? Why did you not take my advice and sold your cards if you really own some? There are enough people happy to buy it and use it with what it offers. Should I tell you a secret? Noone with sane mind wants to use 68k linux distros if there is plenty of cheap hardware available. Gunnar already answered that linux is no topic for the team, why do you keep posting the nonsense?
 
  Really I do not get it...


Kolbjørn Barmen
(Needs Verification)
Posts 219/ 2
09 Oct 2017 13:58


Again, this has nothing to with Linux, it has to do with whether Vampire and Apollo Core provides the type of open hardware that we are used to, or if it is some sort of Amiga equivalent to old 68k Macs, where "what goes on" is considered intellectual property. NetBSD on Amiga is also well and alive, are you saying it would be "insane" to run NetBSD on a Vampire system?

And what about the Apollo team? It should be none of their worries if someone wants to implement whatever OS on a Vampire, should it? Unless it somehow _MUST_ involve them.


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
09 Oct 2017 14:16


Kolbjørn Barmen wrote:

@Gunnar
 
  The core aimed for V2 - does it already, or will it have the 68080 MMU?
 
  Will it be possible for anyone with that kind of interest to use that MMU, without having to go through you first?

The MMU is hidden and not accessible for you.
Technically this is for today the only reasonable solution.

The 68080 MMU is by design different to previous 68K MMU.
The 68080 has not only 1 bus - but 2 busses and 2 memory controllers. So part of the MMU is to define which page is on which bus.
Also 68080 is not a 32bit machine but a 64bit machine.
And 68080 support more protection flags. E.g. not only protect from READ, WRITE - but also protect from EXECUTION.
Also new is that 68080 can control burst behavior.

So its crystal clear that these new features can not be provided on  the same interface of e.g. an old 68030 MMU.

If we would "expose" the MMU then your old software would wrong program it - and you would crash or loose data.

As we protect the others from this - the MMU is today kept private.




Nixus Minimax

Posts 416
09 Oct 2017 14:23


Kolbjørn Barmen wrote:

Again, this has nothing to with Linux, it has to do with whether Vampire and Apollo Core provides the type of open hardware that we are used to, or if it is some sort of Amiga equivalent to old 68k Macs, where "what goes on" is considered intellectual property.

When did Motorola ever do public beta testing of their processors? Did they ever publish RFCs for their MMUs which happened to always be incompatible to each other from one processor generation to the next? Motorola processors never were "the type of open hardware that we are used to". Quite the opposite, it was very much closed (albeit documented) hardware. You didn't get any say in which instructions you wanted to have included in hardware or which you wanted emulated in software. And it didn't take any public demands from clever people like you to make Motorola make the 020 compatible to the 000.

You still don't get it. You are given the opportunity to follow the development of a complex CPU. This is the one important difference between the old 68k processors and the new 080.

If something is not publically available NOW, it doesn't mean it will be forever. (Having something now is still better than having nothing).

If something is incompatible NOW, it doesn't mean it will be forever. (Being able to run a large portion of the existing software base is better than running nothing or a smaller portion).

If something is undocumented NOW, it doesn't mean it will stay undocumented and closed forever. (Some information is better than a completely black-box model of development).

If Linux doesn't run NOW, it doesn't mean it won't ever. (Linux required changes to run on AMD64, ARM changed the program counter register from a combined register with 26bit PC and processor status flags to a 32bit PC register necessitating changes to all ARM operating systems).



Przemyslaw Tkaczyk

Posts 155
09 Oct 2017 15:00


Nixus Minimax wrote:

Kolbjørn Barmen wrote:

(...) it has to do with whether Vampire and Apollo Core provides the type of open hardware that we are used to (...)
 

 
  When did Motorola ever do public beta testing of their processors? Did they ever publish RFCs for their MMUs which happened to always be incompatible to each other from one processor generation to the next? Motorola processors never were "the type of open hardware that we are used to". (...)

Exactly what I think. Thank you, Nixus, for summing up what I was trying to explain to naysayers too many fkin times already. I feel it's one of the greatest developments in Amiga history and being a part of it is a great priviledge.


Kolbjørn Barmen
(Needs Verification)
Posts 219/ 2
09 Oct 2017 16:03


Gunnar von Boehn wrote:

  The MMU is hidden and not accessible for you.
  Technically this is for today the only reasonable solution.

Yes, exactly my understanding.

If we would "expose" the MMU then your old software would wrong program it - and you would crash or loose data.

Does this also imply that the MMU is already used by the firmware that comes with Vampire cards?

posts 158page  1 2 3 4 5 6 7 8