@Arko

Quote:

Arko wrote: @megol Quote: megol wrote: Supporting big endian code on x86 is easy, have been easy and have been done a number of times. In fact there have been a JIT 68k40 -> x86 available on the market before AOS4 was conceived (don't remember the name ATM). JIT is not a problem solver, it is just an emulation technology. It might run some things faster but i won't make them run better.

Hmm? Yes it does as the JIT system I refer to (not by name sadly, still can't remember the name - syn* something?) used the JIT to avoid some runtime byte swaps. However the reason I mentioned it was to establish that before AOS4 coding began there already was a proven solution for running 68k code on x86!

Quote:

The main problem is somewhere else: 68k programs are accessing the system structures in 68k byte order. x86 programs are accessing the system structures in a x86 byte order. So both system can't access the same system structures. Even the emulator doesn't know when a 68k structure will be needed in an x86 byte order.

The solution is simple: just let x86 code run in big endian "mode". That's what Amithlon did and it can be done faster as I have explained on a number of sites.

Quote:

Well you can't change the programming API on 68k software because most of it is not supported any more. So you would need some kind of sandboxing more complex than the 68k emulators on UAE, MOS or AOS4 does. ...

No. The first version would need no sandboxing at all, just create a virtual big endian x86 environment and use a JIT/AOT solution for running 68k code. Native x86 code could still replace parts of the system just as Amithlon supported.
A later version would need a sandbox solution for providing modern features like memory protection and SMP support and then I'd think it be reasonable to have a native code subsystem and a compatibility subsystem.

Quote:

... And even Rosetta on OSX did not provide a full transparent emulation, it could not run 68k drivers for scanners on x86.

That's a choice they made and a reasonable one. There's nothing that makes that the only one possible though...

Quote:

But there might be a solution for this problem, you should aks how many old programs are used that can't work under UAE. For example JUAE on AROS might be a good example how the 68k support could be done today. It will not support 68k drivers for scanners over the emulation but it will support most 68k applications and games.

It is a solution but a pretty bad one. I'd expect a modern AmigaOS to offer much tighter integration between the subsystems than a completely separate emulation. Hardware hitting games/demos could still use such a solution though.

Anyone wanting to do an unofficial "x86" port (AMD64 is the only reasonable target now) could do worse than targeting AMD E1-3000 or other "Kabini" chips.
They support the MOVBE instruction which makes emulating big endian operations very easy. They have a reasonable powerful GPU. They are low priced. They are a SoC solution which makes it easier to fully support the included hardware. They even support virtualization so that it could be possible to run Windows or Linux in a AmigaOS window :)

@megol

I'm thinking if it was possible to define types as big endian or little endian then it should be possible to generate code that uses the right assembler instructions.

For example:

unsigned bigendian int x;
unsigned littleendian int y;
int bigendian *ptr;

if this was possible then assembler code then a lot of the problems between two CPU types might been less, just imagine how this can dis complicate things.

Now you need to use macros that are often not assembler optimized, or inline assembler to solve this.

Last edited by NutsAboutAmiga on 17-Mar-2013 at 04:55 PM.
Last edited by NutsAboutAmiga on 17-Mar-2013 at 04:53 PM.
Last edited by NutsAboutAmiga on 17-Mar-2013 at 04:45 PM.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS

Only meaningfull post in this thread, wich contibute something meaningfull is what Franko wrote:

http://amigaworld.net/modules/newbb/viewtopic.php?mode=viewtopic&topic_id=37413&forum=33&start=60&viewmode=flat&order=0#701414

Personally I'm sick and tired this meaningless over and over again thing, about this subject.

Last edited by utri007 on 17-Mar-2013 at 05:16 PM.
Last edited by utri007 on 17-Mar-2013 at 05:15 PM.
Last edited by utri007 on 17-Mar-2013 at 05:15 PM.

@royleith

x86 is nothing more than a power hogging (hence battery eating) over-glorified calculator CPU. Oh and I'm actually typing this on a Mac that is powered by Intel, so I'm dogging my own hardware

Also, though PPC was better than x86, Intel did a WAY better job of pushing the envelope and sticking to its guns while PPC was shifted to focus on the embedded market.

I would love to see Amiga on ARM!

Last edited by AmigaMac on 17-Mar-2013 at 05:19 PM.

_________________

@AmigaMac

Quote:

x86 is nothing more than a power hogging (hence battery eating) over-glorified calculator CPU.

Bollocks :D

It's probably the most efficient ISA out there, perf/Watt. Not counting Bulldozer :D

_________________

@paolone

Quote:

Yes, there aren't desktop computers with ARM (or, more precisely, there are some netboks but their numbers are scarce). Have you ever wondered why?

We know why their numbers are scarce. Microsoft put restrictions on Atom based netbooks (10" screen maximums, etc) to ensure that the market for their high priced Windows OS was maintained. The Wintel alliance wanted to keep out the ARM chip from the netbook because it would promote Android/Linux. Also, the better battery life of the ARM based systems militated against the Intel Atom, and still does even with the latest versions.

To see the challenge, compare the size and battery life of the Microsoft Surface RT and Surface Pro.

Then look at the performance of the iPad and Android touchpads.

The performance of the Raspberry Pi with linux (single core, 700Mbit, 32bit) is good. It is nowhere near as powerful as the latest multicore, graphics core ARM chips which would make a great budget Linux desktop. The 64bit generation of ARM would be even better.

Porting AmigaOS to the Raspberry Pi would be a great proof of concept and could pay for itself with large volumes of OS sales at a reasonable price (wouldn't you be prepared to double the cost of your Raspi computer in order to run native AmigaOS?). It would have the advantage of driver code availability, 1080p graphics, fixed hardware platform and an inability to challenge the market for a full desktop, high performance ARM motherboard.

Roy Leith

Quote:

AmigaMac wrote: @royleith x86 is nothing more than a power hogging (hence battery eating) over-glorified calculator CPU. Oh and I'm actually typing this on a Mac that is powered by Intel, so I'm dogging my own hardware

X86 is the most effective for medium performance processors (=high end workstations) and even some super computer tasks. It isn't the best for low power tasks though.
And the calculator thingy can't be more wrong. The Intel 4004 was a calculator chip and the 8086 and descendants was designed for high level languages and generic processing tasks. They aren't related.

Quote:

Also, though PPC was better than x86, Intel did a WAY better job of pushing the envelope and sticking to its guns while PPC was shifted to focus on the embedded market.

Better how? Altivec was better than early SSE versions I grant you that but PPC have no genuine advantages architecturally IMHO. For a RISC design it is complicated and high performance designs need to do similar things that x86 does (splitting instructions ++).

Quote:

I would love to see Amiga on ARM!

Who wouldn't?

Quote:

I would love to see Amiga on ARM!

Here you go

@resle

LOL

@vision

Quote:

The future is multi-device and tactile

Imagine typing code in Eclipse with one of that ridiculous and uncomfortable screen keyboards? no thanks. I already don't like laptop keyboards so that would only make it worse.

Quote:

it is what the entire world already chose.

I don't give a f*** about what the entire world chose. Perhaps you didn't notice it but nowadays a computer is like a microwave oven or a freezer. I try to avoid following trends and I don't care much about what people with zero interest in computers do about computers. Now the trend is playing angry birds and using wassap, line or whatever s**t is in awe. I don't care.

Quote:

there are driver restrictions but A LOT LESS than in the x86 world

LOL. There are not many things better documented than x86 hardware.

Quote:

video playing (you can play 1080p movies with devices for under 60$. How much does that cost for a complete x86 system?

Are you sure? Without using binary blobs? My raspberry Pi can't without using linux binaries. It already does his job pretty well: xbmc.

Quote:

MOBILE DEVICES TOUCH SCREENS

Perhaps your problem is that you want that an obsolete desktop OS evolves magically into some kind of mobile phone OS just because you have some kind of brand-nostalgia. People don't care much about the OS of their mobile phones, they just want the trendy apps and that's all.

_________________
The only spanish amiga news web page/club: CUAZ

@moderators

Can we please just lock this?! And post a locked sticky topic that says "No 'processor wars' and no 'port to this hardware' threads"?!

In that locked thread the mods could provide links about the official status regarding the intentions of the development teams and update whenever these intentions change. I am not sure myself if for example, When a development team can to port to a hardware, If a development team even wants to port to a hardware, and Why.

Posting these threads doesn't seem to be a solution so instead ... If anybody wonders if a development team is ready to port their OS to a hardware, they can always check the locked sticky thread for updates.

Last edited by In_Correct on 18-Mar-2013 at 06:26 PM.

_________________
BoingBlogs Wiki (under construction)

AmigaOS = MorphOS = AROS = RISC OS

@In_Correct

Quote:

n that locked thread the mods could provide links about the official status regarding

etc.

We already have that without requiring links. That's what this forum category is for, and the sticky makes that clear.

example

#6

_________________
This posting, in its entirety, represents solely the perspective of the author.
*Secrecy has served us so well*

Could this thread be moved away from OS4 hardware? At this point of view this is totally off topic.

@resle

You deserve a refund from your tattoo artist! LOL

@ferrels

Just found it online, I wonder who the guy is.

But - you know, that tattoo makes me think. We are talking hw architectures here, and yet people can't even settle on a single amiga LOGO...

@royleith

Quote:

We know why their numbers are scarce. Microsoft put restrictions on Atom based netbooks (10" screen maximums, etc) to ensure that the market for their high priced Windows OS was maintained. The Wintel alliance wanted to keep out the ARM chip from the netbook because it would promote Android/Linux. Also, the better battery life of the ARM based systems militated against the Intel Atom, and still does even with the latest versions.

MIcrosoft have absolutely no influence on the size of atom based netbooks. There was also plenty of Linux netbooks using an atom. In fact some of the more successful ones (asus eee pc, acer aspire one) started off using a linux distro (Xubuntu iirc), and Windows only became an option later on.
In regards to power consumption x86 is far fom the power consumption hog it used to be. On the desktop it's untouchable (Intel at least, and AMD is making improvements too, albiet a little more slowly), and even at the lower power segment its becoming pretty competitive.

Not that I really care, Im interested in hardware in general, so have no particular "allegiance", but that above paragraph read like an advertisement for, "modern x86 misinformation 101" :)

@royleith

Ben Hermans (Managing Partner at Hyperion Entertainment) is a lawyer, and after getting his way in the mediated lawsuit with Amiga Inc. I'm sure he thinks of himself as a good lawyer.

Which architecture should AmigaOS ultimately live on?

Ben should create a public licence model for AmigaOS like GPL or Apache License, under which Hyperion would give limited rights to the AmigaOS 4.x source code and then let the community of developers work on the various possibilities. A bunch will work on an ARM port and another bunch will work on an x86 port. In 12 months we'd be able to see who makes the most progress.

Under the terms of the AOPL Hyperion can take the most progressed work and roll it into AmigaOS 5 with a bunch of Hyperion's own 1st party developments. They keep the AmigaOS 4.x open under the AOPL and see what happens in the next 12 months.

It can't be much worse than what they're doing now.

_________________
All the way, with 68k

@royleith

Not going to read through the thread, so just my comment:

"Since the dramatic increase of the X1000 PPC chip prices, I think we need to explore the alternatives. "

x1000 was a special case. It was based on CPU that was rare and no longer in production.

"At the moment, the dominant processor architecture is ARM."

Most units sold, perhaps. The biggest CPU core business is x86, the second is PPC.

" There are already four-core plus graphics chips in general use."

ARM chips are inferior to PPC when considering desktop use.
Built in GPU is not enough for mid or high end desktop.

"It brings with it a wide range of open source software and device drivers."

Less than with PPC.
There is no drivers for high end GPUs, as an example.

There is room for cheap PPC computers for hobbyist.
PPC @1Ghz cost 30 eur, it's not expensive.
One can build a PPC computer with Radeon HD graphics below 200eur.

Other than that. Lets see how AROS does on ARM.

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@agami

Quote:

Ben should create a public licence model for AmigaOS

I doubt he has the right to do so.