osprey atari coldfire + arm still in development

[lp]

It comes down to who is going to do it. Also will you get enough people to use it, to make it worth while to invest the time. I think you'll be hard pressed to find enough people willing to go that route. Start a new thread with a poll.

I won't be jumping on the intel bandwagon. Comparing a dead platform to live ones is rather skewed in my opinion. This platform is not bound to any such requirements.

Now on the other hand if you are trying to say this osprey is a corny idea, that much I agree on.

[Desty]

Apple switched over to Intel because they wanted to keep selling desktop computers that would compete with PCs.
I don't think anyone here would kid themselves that somehow remnants of the ST line can compete with PCs if you shoehorn it into an Intel architecture?
If you lose the hardware, there's just the OS. What's that? Not much more than a copy of CP/M with some nice VDI->AES->GEM layers on top. If you're really that pushed about it, there's already OpenGEM, although it's obviously a lot different than what's on our Ataris.

Personally, I don't need to resurrect the ghost of my ST onto PC architecture - it wouldn't be backward-compatible anyway, and there's nothing amazingly great about VDI, AES and GEM in any case. I'm happy with my STe and Falcon sitting upstairs just for fun, and my Macbook downstairs (for fun and non-fun). And my PS2... and Tungsten T3.

In this scenario it will make it imminently more practical to use Atari software because it overcomes the problem of having to own more than one computer if you want to be active with Atari

That's not a problem! I'm sure almost everyone here is happy to have a modern computer and an Atari. If you offered to take away their old, battered, slow ST and shiny PC and give them some PC-based hybrid "Atari", I don't think anyone would be interested - honestly. For people who really want 1 computer, then they will just use an emulator for Atari stuff. Even I use Hatari much more than I go upstairs to boot up the STe and Falcon (actually, I haven't switched on the Falcon for 6 months or so, due to the crappy setup in the room).

[Omikronman]

"Apple switched over to Intel because they wanted to keep selling desktop computers that would compete with PCs.""

No. Apple switched to intel CPU´s because Intel was not able to offer a mobile G5 CPU for new Apple Notebooks.

[rocket-dog]

Our Celtic friend has a point. I want to start using Atari hardware more for the experience (whirring discs, lovely keyboards, authenticity etc.) I might do simple text input and I intend to hang them of Linux boxes as terminals. But they can't replace modern kit which through technical progress are far in advance of Atari's. (Though I have hardly seen anything new in IT for nearly decade.) I am interested though getting a small single board computer (such as the BeagleBoard) to use as a host for ST emulators. At a £100 or so I would be buying new kit that will give me any configuration of ST I want. A much better bet than buying £200+ of Falcon, TT or what have you.

[jkolak]

Apple switched over to Intel because they wanted to keep selling desktop computers that would compete with PCs.
I don't think anyone here would kid themselves that somehow remnants of the ST line can compete with PCs if you shoehorn it into an Intel architecture?
If you lose the hardware, there's just the OS. What's that? Not much more than a copy of CP/M with some nice VDI->AES->GEM layers on top. If you're really that pushed about it, there's already OpenGEM, although it's obviously a lot different than what's on our Ataris.

Personally, I don't need to resurrect the ghost of my ST onto PC architecture - it wouldn't be backward-compatible anyway, and there's nothing amazingly great about VDI, AES and GEM in any case. I'm happy with my STe and Falcon sitting upstairs just for fun, and my Macbook downstairs (for fun and non-fun). And my PS2... and Tungsten T3.

That's not a problem! I'm sure almost everyone here is happy to have a modern computer and an Atari. If you offered to take away their old, battered, slow ST and shiny PC and give them some PC-based hybrid "Atari", I don't think anyone would be interested - honestly. For people who really want 1 computer, then they will just use an emulator for Atari stuff. Even I use Hatari much more than I go upstairs to boot up the STe and Falcon (actually, I haven't switched on the Falcon for 6 months or so, due to the crappy setup in the room).

It's not to compete with modern the modern PC, but rather that the modern PC will run any software from any hardware platform via virtual machines. Rather than putting ST chips in FPGAs, they will be in virtual machines. Maybe instead of the VM model we will see JAVA type VMs that are truly platform independent. Our beloved Ataris are just a collection of logic gate circuits, and as the machines decay, they can live in virtual code.

[alexh]

Apple switched to intel CPU´s because Intel was not able to offer a mobile G5 CPU for new Apple Notebooks.

Freescale even!

Anyhow you're both wrong. Apple started work on the Intel platform when they commissioned MacOS X. All versions of MacOS X (including pre-release betas) had been compiled and run on Intel hardware.

[lp]

To jkolak:

But your plan fails to cover backward compatibility. To this day there is no emulator that fulls off running full fledged Falcon software with DSP at any acceptable level. So everyone would still have their old machines up stairs. So it solves nothing in the end. Trying to convince everyone they should throw out there old wares is an even tougher sell.

[Desty]

Apple switched to intel CPU´s because Intel was not able to offer a mobile G5 CPU for new Apple Notebooks.

Anyhow you're both wrong. Apple started work on the Intel platform when they commissioned MacOS X. All versions of MacOS X (including pre-release betas) had been compiled and run on Intel hardware.

Yes, but that's not why they switched to Intel. They started that work some years earlier, just to have the option if they needed it.

The main reason was indeed that IBM were failing to improve the PowerPC architecture, at least not enough to keep up with Intel.
And thus, the Mac wouldn't be able to compete with PCs unless they (as they saw it) they switched to Intel.
Not that there weren't other options... I'm personally looking forward to the day we move onto a more powerful architecture than the baggage-encumbered x86.

[jkolak]

To jkolak:

But your plan fails to cover backward compatibility. To this day there is no emulator that fulls off running full fledged Falcon software with DSP at any acceptable level. So everyone would still have their old machines up stairs. So it solves nothing in the end. Trying to convince everyone they should throw out there old wares is an even tougher sell.

Hi,

It's not really about anyone giving up their hardware. It's about preserving this portion of computing history.

As this thread shows,

viewtopic.php?f=15&t=15542

15 to 20 years are at the outer limits of circuit board life. Taking the long view, how many STs will be around in 2030? And how many will you be able to find outside of museums and private collections in 2050? It's 90 years since the end of WWI. How many Sopwith Camels and Fokker triplanes are still available to look at, let alone fly? How many STs will be running in 50 or 100 years? In order not to lose this piece of history, computer historians with technical skills will need to locate chip schematics so all hardware and software can be reproduced in RAM, which will give each machine variation perfect compatibility. A boot menu might offer you which machine configuration you want to run. In this way the experience of the Atari ST will be available in software for future generations. This path works for the obsolete equipment. For the path to the future of TOS computing, it looks like AFROS is the direction things are going now. In the interview over at XaAES

http://xaaes.atariforge.net/index.php?s ... &i=ozk.htm

Odd Skancke talks about taking the machine dependencies out of Mint. Again, this is not to take away anyone's hardware, but rather to let people build their TOS computers with commonly available mass market components instead of waiting for another clone maker.

John

[jkolak]

Our Celtic friend has a point. I want to start using Atari hardware more for the experience (whirring discs, lovely keyboards, authenticity etc.)

That's one thing I loved and missed too -- Atari made beautiful computers. They hired an Italian sculptor for the ST case. This worked for VW too on some of their more striking car designs. The Ghia and Jetta come to mind off hand. I put a line or too about this in the wikipedia article on the ST.

[lp]

Odd Skancke talks about taking the machine dependencies out of Mint. Again, this is not to take away anyone's hardware, but rather to let people build their TOS computers with commonly available mass market components instead of waiting for another clone maker.

Odd had a PPC dev board shipped to him. Hence his comments. Pretty sure he's not on the x86 band wagon either. I have to agree with Desty, something better needs to come along. That may spark more interest.

[rocket-dog]

That's one thing I loved and missed too -- Atari made beautiful computers. They hired an Italian sculptor for the ST case.

I didn't know that!

I know the case design of a machine don't determine performance but it is another dimension to the hobby. I would like a Sinclair QL purely for its blackness.......

[jkolak]

Odd Skancke talks about taking the machine dependencies out of Mint. Again, this is not to take away anyone's hardware, but rather to let people build their TOS computers with commonly available mass market components instead of waiting for another clone maker.

Odd had a PPC dev board shipped to him. Hence his comments. Pretty sure he's not on the x86 band wagon either. I have to agree with Desty, something better needs to come along. That may spark more interest.

Darek Mihoka's No Execute! blog at emulators.com talks alot about CPU issues and problems blended with his Atari, Mac, and Microsoft experience. He felt Intel was on the right track with the Itanium CPU and that it was a mistake when they backed away from that.

[jkolak]

I know the case design of a machine don't determine performance but it is another dimension to the hobby. I would like a Sinclair QL purely for its blackness.......

I liked Rakoon's idea of the black TT. The "ugly" STE on ebay right now would have been okay if the stone paint texture wasn't used. I once thought stone textures were nice for VW buses for a different look, but on a computer it pretty much does away with the chance to change the color again. The Falcon prototypes were black. The Jaguar looks good in black too.

[Shredder11]

I know the case design of a machine don't determine performance but it is another dimension to the hobby. I would like a Sinclair QL purely for its blackness.......

Yeah I remember when the QL came out and at the time, the nearest I got to it was having my Spectrum 48K rubber key model upgraded to the 48K Plus under guarantee (had 3 or 4 go faulty until no more rubber models were available for me). I always thought those Microdrives looked really cool and cute back then too, although history has shown how unreliable they were.

At some point in the future I fancy buying some more old micros, such as the ZX80/81, Spectrum 48K amongst others. Just need to move house and find more space!

[Desty]

Darek Mihoka's No Execute! blog at emulators.com talks alot about CPU issues and problems blended with his Atari, Mac, and Microsoft experience. He felt Intel was on the right track with the Itanium CPU and that it was a mistake when they backed away from that.

Hmm... it was nice to see Intel openly explore other avenues than the smelly x86, but there were a lot of valid criticisms of Itanium - it was strongly focused on ILP, downsizing conventional branch-prediction logic and out-of-order execution circuitry in favour of relatively gigantic caches, needing more transistors for cache on the die to support core logic and affecting scalability.
Also, the (extremely) VLIW instruction set was possibly expecting too much from compiler writers (and probably made hand-coding in assembly almost impossible ). In fact, Donald Knuth even said "was supposed to be so terrific—until it turned out that the wished-for compilers were basically impossible to write".

[jkolak]

Hmm... it was nice to see Intel openly explore other avenues than the smelly x86, but there were a lot of valid criticisms of Itanium - it was strongly focused on ILP, downsizing conventional branch-prediction logic and out-of-order execution circuitry in favour of relatively gigantic caches, needing more transistors for cache on the die to support core logic and affecting scalability.
Also, the (extremely) VLIW instruction set was possibly expecting too much from compiler writers (and probably made hand-coding in assembly almost impossible ). In fact, Donald Knuth even said "was supposed to be so terrific—until it turned out that the wished-for compilers were basically impossible to write".

Thanks Desty. That was good information. What CPU do you think is the best now?

[Desty]

Thanks Desty. That was good information. What CPU do you think is the best now?

It's hard to say really - different architectures suit different purposes and exploit different strengths. I wanted to believe the promise of ILP and massively multicore chips, but so far it just hasn't panned out all that well due to the enormous software complexity involved in current software approaches to parallelism (i.e. the way things are today, having a million cores won't help at all if your code/compilers can't exploit it). Intel has done a pretty good job at keeping the x86 competitive somehow, even with its inherent drawbacks, too. The last really interesting architecture I saw was the Cell, but again, its particular advantages only help on some kinds of problems that can be parallelised (e.g. cryptography, modelling, games, graphics).
For mobile devices though, none of that really matters much at the moment - lots of registers and a decent RISC instruction set and cache are enough (RISC -> less complex circuitry required -> less heat generated/wasted -> less power used) - which is why ARM is still the king there.

If we start to see some progress on the software side of massive-multithreading/ILP, then the multicore world will be more promising (have you seen the CUDA parallel stuff on NVidia graphics cards?).

[jkolak]

It's hard to say really - different architectures suit different purposes and exploit different strengths. I wanted to believe the promise of ILP and massively multicore chips, but so far it just hasn't panned out all that well due to the enormous software complexity involved in current software approaches to parallelism (i.e. the way things are today, having a million cores won't help at all if your code/compilers can't exploit it).

I see that all the time in my quad core. When I set up a task that takes an hour on a single core, instead of it finishing in 15 minutes at 100% CPU usage, it still takes an hour and runs at 25% CPU usage. In fact, I rarely ever see my CPU usage meter go over 25%.

Intel has done a pretty good job at keeping the x86 competitive somehow, even with its inherent drawbacks, too.

Taking a time slice snapshot of the Atari ST's moment in time, we chose our STs at a time when Intel chips were badly crippled by backwards compatibility with poor design choices in the 8088 chip. With the 286 and even 386 chips choosing to perpetuate that backwards compatibility, it left the Motorola 68000 chip at a great advantage and we enjoyed that performance advantage in the Atari ST, coupled with other design efficiencies in the ST which really made it a quick and snappy computer to work with in the early days when programs were small.

When Bill Gates was once criticized for Windows being slow because the code is inefficient his response, instead of saying they would improve the code, was that advances in CPU speed would take up the slack. Well, the truth is that Windows is not more responsive because the bloat has probably exceeded the CPU speed increase such that todays Wintel computers really do not feel faster than they did in the 80's. But from an Atari users perspective, who would have thought that we would see 2 orders of magnitude increases in CPU speed. It wasn't long before 33 MHz turned into 300 MHz, and now 20 years later we are seeing 3 GHz. Even though Intel is still carrying the backwards compatible legacy core code, we just can't touch the sheer speed their processors are giving. In spite of complaints of successive Windows versions not running all of people's old programs, from what I have read, Windows is also somewhat crippled by carrying legacy code forward into today's versions, and Jerry Pournelle's column wrote on wishing that Microsoft would dump Windows core code for something modern.

If we start to see some progress on the software side of massive-multithreading/ILP, then the multicore world will be more promising (have you seen the CUDA parallel stuff on NVidia graphics cards?).

The article and chart at that link are very interesting. I see I missed the boat with CUDA architecture starting in GeForce series 8 cards. (I have a series 7 on my MB).

[TTowner]

I have owned Atari computers from the day get go.
I like other users have machines that are excellent at what they do.

I recently added to that colection a Hades060. This for me was the pinnacle of my dreams. I hoped that not only would it run most if not all the sofware I have collected over the years. Its capability is far better than the standard ST, yet it works with the standard ST and its software, The next step for me would to some be silly, I would like to take advantage of any improcements of the machine both hardware and software wise.
As the hardware side is limited, and not a lot is available, it would be usefull to me personally to have some software that would seriously improve my chances of using the Hades to it's full potential.
Any mix of software that would do this is desireable. The Coldfire project is watched as much as I can find, I watch because there is hope that the new hardware will co-operate with existing software so much that I would be more than wwilling to save the money to purchase one.

Having that kind of support from existing users can only benefit the modern Atari And clones.

This discussion, as much as read by me, is facinating,I am deeply interested to see how things pan out.

I have some posts on the forum about trying to install mint on my machines. From a users point of view, it is a nightmare, there are so many different software improvements such as netbsd, aranym, mint, debian that I am totally confused as to what to install. Now, Add the fact that what some people say is easy to install.

A simplified setup, or a complex one? My point is this, I like most others perhaps, are reliant upon people that provide upgraded software, new developed software, to offer me a choice and a path to follow.
Having petty discussions is something that is not productive to me. Statements like Oh one click on mint install and it will install aand work right out of the box. is both none helpfull and misleading.
Sure you are all talking about a future development of a systems functionable software, please consider whilst making the choice to support the toaster or washing machine or the cpu, can you take into account that a one click install of anything you ultimately produce does not necessarily mean that it will suit all custom installations.!
Mint is something I aspired to, simply because it offered, or so I thought, the opportunity to multitask, like a PC.

I couldn't wait to get it and install it.

Asking around for guidance, I got some comments that quite franky are ludicrous to say the least.
If I was to follow them, I would find that the installation would end up useless to my needs.Let alone that I would not get an install at all.
Mint one click install is all well and good. A little more selectivity on what will be installed along with set zip files of componentised addons would have been more usefull. :- Graphics, sound, networking modules would just have a better feel to installing what is prefered.Even if it was essential, user participation would engage the users understanding of the consolidated parts.
So, now, I am holding off installing any suplimentary opperating system, simply to see if there will be anything more user friendly.Because I am buggered if I know what to install

A lesson for you programmers..... What happens when your software becomes so simple for YOU to install, yet the user remains baffeled by the simplest of choices in it's installation...?
Think about it for a moment.
I would like to make the above comment stick in your minds. I may well be off topic, but I believe my point is relevant!

Thank you for your time reading this..

[rocket-dog]

One of things that amazes me about software over a half-century into the computer revolution is how crap most software is from the UI perspective (this painfully evident in the design of most web pages.) Especially when you consider how much research has gone on into that area.

[joska]

If I was to follow them, I would find that the installation would end up useless to my needs.Let alone that I would not get an install at all.
Mint one click install is all well and good. A little more selectivity on what will be installed along with set zip files of componentised addons would have been more usefull. :- Graphics, sound, networking modules would just have a better feel to installing what is prefered.Even if it was essential, user participation would engage the users understanding of the consolidated parts.
So, now, I am holding off installing any suplimentary opperating system, simply to see if there will be anything more user friendly.Because I am buggered if I know what to install

You have been advised many times to try EasyMiNT. Have you done so?

[wongck]

If I was to follow them, I would find that the installation would end up useless to my needs.Let alone that I would not get an install at all.
Mint one click install is all well and good. A little more selectivity on what will be installed along with set zip files of componentised addons would have been more usefull. :- Graphics, sound, networking modules would just have a better feel to installing what is prefered.Even if it was essential, user participation would engage the users understanding of the consolidated parts.
So, now, I am holding off installing any suplimentary opperating system, simply to see if there will be anything more user friendly.Because I am buggered if I know what to install

Wow.... may be OS2 will install in 1 click!!!

[TTowner]

If I was to follow them, I would find that the installation would end up useless to my needs.Let alone that I would not get an install at all.
Mint one click install is all well and good. A little more selectivity on what will be installed along with set zip files of componentised addons would have been more usefull. :- Graphics, sound, networking modules would just have a better feel to installing what is prefered.Even if it was essential, user participation would engage the users understanding of the consolidated parts.
So, now, I am holding off installing any suplimentary opperating system, simply to see if there will be anything more user friendly.Because I am buggered if I know what to install

You have been advised many times to try EasyMiNT. Have you done so?

Yes, I have tried easy mint..Have tried several times.
Even on a simple install. No extras. I even tried cut down versions but that does nothing to help.

[joska]

Yes, I have tried easy mint..Have tried several times.
Even on a simple install. No extras. I even tried cut down versions but that does nothing to help.

What goes wrong? Have you informed the creator of EasyMiNT? Even on strange hardware like my Milan060 (of which almost none were sold) EasyMiNT installed without any problems whatsoever. I would be very surprised if it didn't work on a plain TT.