Ideas: The new Amiga. Yes, it's inevitable!

[persia]

I would like it pocket size with a direct mind input/output system.  Powered by cold fusion.

[alexh]

@ Persia
We've already had both of those.

But I just realised that a direct mind I/O system would require all Amiga users to have minds :-D

[Atheist]

bloodline wrote:

But some people here are not expressing realistic desires for a new computer... let alone an amiga...

Hi bloodline,

Well, at least the one I commented on will be made, so I'm happy.

Okay, so they might not go the 4 extra GPUs route to get SLI mode, but it could be made into a 900 MHz ASIC in the future, with enough $$$.

A SUPER Amiga is coming next year, 2009!

I am referring to a >600 MHz ASIC of the SuperAGA.


DISCLAIMER: I am NOT saying NatAmi group of devs is making a >600MHz ASIC, but that it COULD happen.

[alexh]

And I COULD win the Euro-Millions lottery (if I actually played) the odds are probably better.

[dammy]

I think it's amusing how people like you feel they must contribute to the forums, yet you have nothing to contribute. This thread is meant only for fun and conjecture--you may leave at any time.  

Actually, I am making a contribution to this thread, cold hard reality.  Now, isn't that a refreshing injection into a rather beaten to death :horse: subject?

Dammy

[dammy]

I would strongly argue there. ARM is not the only real answer, even among portable makers. The Intel Atom, VIA Nano both make solid x86-capable portable CPU's which are making inroads. New MIPS CPU's are arriving that put the ARM in for a run for it's money. Even the ancient 68k with ColdFire is starting to bring some excitement back to the field. ARM might be 60% of the market, but that is not the "only real answer" any more than m68k was in the mid 1980's, where it was in the top half and Intel was the underdog.

Last I saw, ARM was closer to 70%, but even at 60% it makes it a no brainer to port to ARM ATPIT.  If MIPS or any other CPU gets 40% or better of the portable market, I'll be happy to see a port to that as well.  Right now, it would be rather stupid to place any significant developer time on anything other then ARM. No one can  afford another PPC like fiasco.

Dammy

[T3000]

ZeBeeDee wrote:
Can't do that ... I'm using the processor as a BBQ  :lol:

That would be pan frying.

Let's face it, Microsoft is too powerfull.
[color=008000]All bow down to the almighty Bill Gates.[/color]

[SamuraiCrow]

dammy wrote:

I would strongly argue there. ARM is not the only real answer, even among portable makers. The Intel Atom, VIA Nano both make solid x86-capable portable CPU's which are making inroads. New MIPS CPU's are arriving that put the ARM in for a run for it's money. Even the ancient 68k with ColdFire is starting to bring some excitement back to the field. ARM might be 60% of the market, but that is not the "only real answer" any more than m68k was in the mid 1980's, where it was in the top half and Intel was the underdog.

Last I saw, ARM was closer to 70%, but even at 60% it makes it a no brainer to port to ARM ATPIT.  If MIPS or any other CPU gets 40% or better of the portable market, I'll be happy to see a port to that as well.  Right now, it would be rather stupid to place any significant developer time on anything other then ARM. No one can  afford another PPC like fiasco.

Since nobody can afford another PowerPC-like fiasco then maybe we should be porting to a bitcode that compiles at install time like LLVM.  That's what Apple is doing.  In fact Apple is including LLVM in its 3.1 XCode presumably so it won't ever get stuck on one processor again.  (The fact that they're doing this now that they are on Intel's chips also seems to have escaped you.)  This way they can have Intel chips and PowerPC chips on the desktop and ARM chips on the handhelds without having to widen their universal "fat" binaries to produce all kinds of code simultaneously at compile time.

[dammy]

Since nobody can afford another PowerPC-like fiasco then maybe we should be porting to a bitcode that compiles at install time like LLVM. That's what Apple is doing. In fact Apple is including LLVM in its 3.1 XCode presumably so it won't ever get stuck on one processor again. (The fact that they're doing this now that they are on Intel's chips also seems to have escaped you.) This way they can have Intel chips and PowerPC chips on the desktop and ARM chips on the handhelds without having to widen their universal "fat" binaries to produce all kinds of code simultaneously at compile time.

LLVM gets a thumbs up from me, at least for AROS.

Dammy

[Zekaric]

[replying to the original post]

The mythical Amiga of the future...  I wouldn't mind seeing it happen.  IMO...

1/ Graphics powerhouse.  Possibly.  Real time Raytracing is seriously in beta stage on the PC.  It'll be at least 2 years before you see something in the market and I think that is being optimistic.  Everyone will need to upgrade their video cards and and machines to play with real time raytraced programs.  That will be a hard pill to swallow for many.  

Compared to the current 3D rendering, raytracing is super computationally intensive.  You need some beefy hardware to support it.  I'd be content with a more normal and current board from nVidia.

Quaternions is just another tool in the box, it won't replace euclidean transformations; so they aren't exactly necessary IMO.

Personally, I'd like to see a good OpenGL implementation instead of a roll and Amiga flavour of a 3D library.  This will make porting 3D games and programs over quite a bit easier and make the Amiga a platform that serious developers might consider.

2/ Developer Support.  No argument here.  I do suggest being ruthless in native library functions.  I don't know enough about the Amiga libraries but I know enough about the Windows libraries to say that functions should be simple and easy to understand and use.  Trim the unnecessary fat.  Make them work logically.  Make them work.

3/ Multimedia platform.  Amiga could have had this market tied up back in the Babylon 5 days.  If only...  But it's one of those niches that a computer needs to carve out for itself.  It will be difficult to claw that back from Windows, Mac and Linux.  Also with all these new video formats you need some beefy hardware just for encoding the video.  You also need, for serious work, RAID support and very large file systems.  

Audio production was also somewhat a field that the Amiga could have dominated.  Also a field that will require some clawing back of the other platforms.  

4/ Hardware Modularity.  I somewhat agree.  Partly what killed or limited the market for SGI, Sun and MAC was the fact that they used proprietary stuff.  And that proprietary stuff had some serious markup.  Users look at the cost of a drive/monitor/keyboard/etc. and look at the similar PC generic device and wonder, what they hell am I paying for.  It was a reason why at my work we dumped SGI and SUN support because the reason to go with them was no longer compelling enough.  PCs were good enough, fast enough and half as costly.  You can buy off the shelf components when certain components failed.  Whatever this mythical Amiga becomes should just use off the shelf components as much as possible.

This brings up an comment I saw someone else make.  Low end should look like an A500/A600/A1200?  What are you smoking?  I never understood why people liked that all in one look.  Keyboards fail.  Replacing a keyboard in an all in one design is far more difficult, especially for the average gumby, than buying an off the shelf keyboard and plugging it in.  I would rather see something like a mac-mini or iTV form factor in a low end device than the computer in the keyboard idea.  

I would ideally like to see digital outs.  HDMI (video and audio) or at the very least DVI with digital coax/optical audio.  The trend is going that way and it would make hooking up the stereo system so much easier.  With 7.1 or more signals and analog out is a bit clumsy.  Multi-head capable is needed as well.  When doing 3D, software dev or maybe even video editing, having multiple heads is a godsend.

I would ideally like to see at least a Ghz machine, but closer to 2 Ghz the better.  There are a few PCs that still come with sub 2GHz range.  Multi-core would be nice but for a first go, not necessary.  64 Bit as well.  When working with big audio and video or even 3D, a lot of companies are hitting the 4GB barrier and are wanting more.  For a serious solution, you need to be able to give it to them.

So as a minimum
Motherboard:
- 64 bit
- GHz level CPU
- >4 GB capable using off the shelf RAM modules
- PCIe expansion capable (if not all in one.)
- Built in video capable of at least 2 heads.  HDMI preferred.
- Built in audio.  Digital coax/optical out preferred.
- USB for perriferals.  (Keyboard, mouse, webcam)
Software:
- Development tools available and free.  (Like Linux and Mac)
- OpenGL libraries.
- Sane OS libraries.
- A nice sane looking interface.  All these 3D enhanced UI that Vista, Mac and some Linux window managers are offering are nice and all but really aren't needed IMO.
- A program that will recognize when I wave a certain finger at the computer that the computer will feel shame.

[downix]

SamuraiCrow wrote:

dammy wrote:

I would strongly argue there. ARM is not the only real answer, even among portable makers. The Intel Atom, VIA Nano both make solid x86-capable portable CPU's which are making inroads. New MIPS CPU's are arriving that put the ARM in for a run for it's money. Even the ancient 68k with ColdFire is starting to bring some excitement back to the field. ARM might be 60% of the market, but that is not the "only real answer" any more than m68k was in the mid 1980's, where it was in the top half and Intel was the underdog.

Last I saw, ARM was closer to 70%, but even at 60% it makes it a no brainer to port to ARM ATPIT.  If MIPS or any other CPU gets 40% or better of the portable market, I'll be happy to see a port to that as well.  Right now, it would be rather stupid to place any significant developer time on anything other then ARM. No one can  afford another PPC like fiasco.

Since nobody can afford another PowerPC-like fiasco then maybe we should be porting to a bitcode that compiles at install time like LLVM.  That's what Apple is doing.  In fact Apple is including LLVM in its 3.1 XCode presumably so it won't ever get stuck on one processor again.  (The fact that they're doing this now that they are on Intel's chips also seems to have escaped you.)  This way they can have Intel chips and PowerPC chips on the desktop and ARM chips on the handhelds without having to widen their universal "fat" binaries to produce all kinds of code simultaneously at compile time.

You got the thumbs up from me here!

[JuvUK]

ok not wishing to start a fight, but why AMD, the old generation i agree kicked intels butt, but the current stuff it's intel all the way, until AMD get up to speed why use them?
 seriously i'm not picking a fight i'm just curious
cheers juv.u.k

[downix]

JuvUK wrote:
ok not wishing to start a fight, but why AMD, the old generation i agree kicked intels butt, but the current stuff it's intel all the way, until AMD get up to speed why use them?
 seriously i'm not picking a fight i'm just curious
cheers juv.u.k

For me it's a case of total solution.  Vendor lock-in is where a single vendor can hold a project hostage, say, Northbridge for PPC chips.  In this case, using the same vendor for as much as possible is a serious discouragement for such tactics.  "We play with their northbridge, wait, they stopped buying our GPU and CPU".  It's a solid tactic.  Intel does not offer this capability, with their sub-par GPU systems.  

AMD also is far friendlier in the licensing needed for producing custom BIOS's, as demonstrated with the LinuxBIOS and OLPC projects.  While they might not offer assistance, they won't hinder either.

Since we lack the ability to manufacture such chips ourselves, the fewer vendors we have to deal with that control critical systems, the fewer chances for some minor supply chain issue to completely derail us.  

[JuvUK]

fair enough, but surely whichever cpu you decide on you are going to end up tied to it in one way or another, and as we've seen, intel is able to weather storms for longer than AMD could hold out, as for a GPU surely that should be a nvidea chip? yet another tie in, but surely the way to go?

[A6000]

The great thing about the amigas custom chipset was/is that it gave each model a long life, 24 years and counting, whilst PCs are obsolete in 4years or less.
If a future amiga used off the shelf parts, it would need to be regularly upgraded too, GPU's have a life span of just 2 years or so, and newer software would not run fast enough on a 4 year old amiga that used off the shelf parts.
Fast processors and gigabytes of ram encourages lazy programming.

[cicero790]

I agree with you A6000. The graphic cards needs to be replaceable other wise it will fall behind pc every second year. It will not do, market wise, and it would fail in the intention of being a computer capable of running the most taxing games available. The option to upgrade must probably be there. The sound chips isn't that sensitive to this rat race or the cpu if its a quality cpu.