Sonnet crescendo 7200 and OS4.1

[Iggy]

ah.. maybe i misunderstood, you mean something like anubis, arix or amithlon? or that kind of amigaos on linux kernel people are fantasizing about? obviously not. otherwise it would be widely known.

you can probably though strip down your linux host to get out of the way as far as possible. except for packages and tools needed for compilation, but then it can all be done from bash shell.

YES, that is exactly what I was thinking! But that could be a future goal. First get it up and running under Debian, and then figure out what can be jettisoned.

Considering the kernel is up to date, it might be possible to keep the underlying distro up to date.

[Iggy]

arent you derailing your own thread now?

Possibly...or I'm deferring to good advice.
You've made some good points, and like I said before, it will be...fun.

And isn't that what hacking is about, challenges, experimentation, pushing the boundaries?

And in the end, we're still discussing PPCs.

Let's just see what kind of port we have ready by the end of the Summer.

[Iggy]

@wawrzon

who is the current maintainer for the PPC platform? if it's possible, i'd love to try booting AROS on my X5K, but am not familiar with doing this with our u-boot firmware. can you recommend anyone?

-- eliyahu

I'd love to hear about that as I intend to buy an X5000, but I'm holding out for the X5000/40.
And my initial interest was in running MorphOS on that, but eventually MorphOS will transition to X64.
Also, I don't have that much interest in Amiga OS4, and AROS did provide some of the initial code to build MorphOS.

Finally, this OS is open and allows the users to participate in its development.
As a hacker since the '70s I find that appealing.

Warzon, eliyahu, lou, again, thanks for the inspiration.
Oh and one more target (two actually) I'd like to look at, iBook and PowerBook laptops.

[Iggy]

@Iggy

if i get it going, i'll post instructions here on amiga.org for others to follow.

-- eliyahu

Cool, um, I'm going to post this on a separate thread, but on the idea of open platforms, Raptor Engineering is planning on producing a reduced cost single cpu version of their Power 9 based TalosII board.
With a four core/sixteen thread processor this board is priced at less than $100 more than an X5000, or in other words its LESS than the price the X1000 sold for.

For a faster system, with a processor that can run either big endian or little endian code (or both) solving our issue with endian problems like the webkit port of Odyssey. And a four core Power 9 cpu can run 16 simultaneous threads, compared to the A1222 or X5000's two (or in the case of the still unavailable X5000/40, four thread, still one fourth at a slower speed).

At a little more than 2 grand with a Radeon HD 5850, some DDR4 and a hard drive, its not cheap, but its not significantly more than Aeon's best.

I might have to buy one and shift to working primarily with Linux hosted AROS.

[Iggy]

I'm not quite sure what to make of that statement. Do you think the other operating system camps don't use version control or something* ?

*that said, the Linux kernel was maintained at one point as a bunch of tarballs. Torvalds invented git because he didn't feel svn was suitable. Having used CVS ans SVN for longer than I can remember, git feels weird to me, but I'm getting there.

I'm not sure I care what Linus thinks, since he obviously added a feature central to micro kernel (loadable/unloadable driver modules) to make up for an obvious dificiency  of monolithic kernels.
I still prefer and support micro kernels and believe them to be supetior.
MorphOS is micro kernel based, the OS I supported on my 68K platforms was micro kernel based, even MacOS is a hybrid, and my eventual goal is to slim down Linux to the minimum needed to run a hosted version of AROS.


You can all go worship at the feet of Torvalds. Because basically Linux was just a kernel designed to facilitate a clone of UNIX. I could probably use BSD instead, but it might be more work.

I'm interested in this because it facilitates what I want to do, not out of some fan boy notion that it's Linux is a sterling example of original thinking.
Come to think of it, nothing in the computing world is. It's all built on other work, cloned copied or outright stolen.
The only thing I've heard of recently that sounded vaguely original was IBM's experiments building neural network based nodes. That, in my opinion, could actually lead to what we need for real AI, since it's similar to how "computing" is done in biologic entities (you know, brains).

In any case, let's freaking do this, let's build on and expand the ATOS PPC base.
Lou got me started thinking about the WiiU, but the X5000 is also a good target. As are the PowerMac and PowerBooks. And the Talos II...that would be incredibly cool.

[Iggy]

Having had access to the source repository in the past, I can confirm that the OS4 sources are maintained under version control. There really is no alternative when working on projects that size.

I was just bemused by Iggy's statement which made it sound like version control was some sort of revelation that only the AROS team used.

Bemused as you may be, I have no interest in OS4, and have no idea how MorphOS developers coordinate their project, although it must be through a similar system.

What amazes me is that allowing contributors to alter existing code doesn't create chaos, obviously there is the option to revert to previous versions, but the loose coordination reminds me a lot of some of the bigger Linux distros.
And often their steps "forward" wipe out something I preferred in earlier revisions.

Obviously there is more centralized control in the OS4 and MorphOS development communities as to what is and what isn't adopted.

In any case, I have the AROS source code, and will move forward in attempting to get it to run stably on some PPC platforms I have an interest in.
lou originally mentioned the WiiU, and I have G4 powered PowerMac, iBook and PowerBook hardware as well as PowerMac G5 systems (both AGP and PCI-e), And while there is a SAM460 port, don't you think there ought to be X1000 and X5000 ports as well?
I'll leave the idea of an A1222 port to those that prefer to punch down instead of up.

Then finally, with the announcement of by Raptor engineering that they will be producing a single cpu variant of the Talos II called the Talos II Lite...

Well we will soon have a Power 9 system available in the same cost range as the X5000.
Think about that, the cheapest Power 9 variant is a quad core that supports 16 concurrent threads.
That's 8 times as many as the X5000/20 and four times as many as the X5000/40.
Heck, its even double the number that the proposed T2080 based laptop could handle.

With a better fpu than any other Power variant thus far.

Porting AROS hosted to a big endian Linux variant on a Power 9 system should not be too difficult.
The next move would be to figure out the changes needed to move it to a little endian Linux variant.
And finally to create a native little endian variant.

And native version of AROS on Power 9 should be able to match any capabilities that AROS has on an X64 system (including the eventual adoption of SMP).
In fact, with the Power 9 hypervisor, it might be a MORE capable system.
Power 9 has already been demo'd running X64 software, and it can run multiple sessions via the hypervisor with sessions being EITHER big endian or little endian.

So think about it, a processor that could run PPC OS4 or MorphOS AND either PPC or X64 AROS (or BOTH), ALL AT THE SAME TIME.
It could even run Windows, Linux or possibly MacOS (on an emulated platform or via a hacked copy of MacOS).

All of this concurrently. In fact, the first four NG OS' mentioned only require 4 thread, leaving 12 threads for Linux or other OS'.

And this is on the cheapest Power 9 cpu.
There are Power 9 variants with up to 22 cores.
Only AMDs Epyc could beat that with 32 cores, but Epyc only supports 2 threads per core versus Power 9's 4 thread per core.

In other words, a single cpu Epyc system could support 64 concurrent threads.
A single cpu Power 9 system could support 24 more, or 88 concurrent threads.
Make these dual cpu systems and that number grows and the Power 9 system can support 48 more threads that the Epyc system.

And here's the kicker, an AMD Desktop Ryzen system maxes out at 16 concurrent threads (the same as the low end Power 9 four core) and can't be upgraded to the enterprise level multi-die Ryzen derivitives or Epyc.

But the Talos II CAN be upgraded. Your can start with the basic four core cpu on a dual or single cpu board (and the dual cpu board can still be used in single cpu configurations).
That gives you the same thread count as the best Ryzen desktop do.
Then you can upgrade all the way up, right to the 22 core cpu if you wish.
Giving you either and 22 core/88 thread machine or a 44 core/176 thread machine.

So, to those of you that contend that the PPC is dead, you have your heads up your rectum (and btw, yes, Power 9 IS a PPC).

This argument traces all that way back to my days building SS-50 bus computers and having arguments with S100 advocates stating their processors were faster.
Sure, and they had a much lower IPC/MIPS count.
And they went on to adopt CP/M (the progenitor of DOS), while we initially settled for Flex (which is pretty much a CP/M clone), but eventually once we had the 6809, we moved to multi-tasking OS' that supported multiple users.

So, the Motorola based community had real priority based preemptive multi-tasking before the Zilog/Intel community came up with kludges like MP/M.

Yes, even the lowly Tandy Color Computer could perform true multi-tasking, and the 68K when introduced was a quantum leap over Zilog and Intel offerings.
In fact, until the i386 was introduced, Intel cpus lacked the features necessary to implement some of the code that was running on the 68K.

SO, hey, the Amiga was the first multimedia computer marketed to the public with a SUPERIOR processor than those used on the first two generations of PCs (and it still easily match Intel's third generation).
The move to Power was intended to provide an alternative to X86, and up to the IBM 970 (aka the G5), it DID have full parity with Intel (in fact in some measures the G5 bests X86 cpus from the same era).

Using PPC as an accelerator for the Amiga was NOT a mistake, it was an effort not retain our exclusivity and not just fall in line with the competition.

If Apple had continued to support IBM/Power, desktop variants of systems above Power4 (the basis for the G5) could have been created that would have maintained that parity.

And IBM has soldiered on on it own, continuing to develop Power, now in its 9th iteration, and its still capable of matching or beating X64.

I'm feed up with having to resort to the "enemies" ISA.
From now on, I'm focusing on ARM and Power.

And if any of you had a clue what IBM has been focusing on as a successor to RISC, you might be question Intel's future dominance as much as I am.
ARM's everywhere in light weight portable devices, and Intel's been giving us more of the same for years.
If it wasn't for AMD we'd probably still be suffering with some lame P4 derivative, and in case anyone has forgotten, the 64 bit extensions to X86 are AMD64 extensions, the X64 ISA was AMD's creation, not Intel's.
They just implemented it as AMD's license requires sharing technology based on Intel underpinnings.

Intel didn't intend X86 to be extended to 64 bit, they wanted to move you all to Itanium. That failed, the P4 was pretty much a failure (both reminding me greatly of the Pentium Pro fiasco).

Hey, remember Intel started with the 4004, which was the first integrated one chip microprocessor, but that development was inevitable and early Intel products sucked.
There were PDP-11 micros produced soon after the 8088/8086 that could run circles around them.
Intel's cpus were best suited to use in  calculators, and you don't even find them there anymore while the Z-80 and 68000 are still used in older TI Scientific calculator designs (and ARM dominates newer designs).

So DO I really HAVE to have an X64 system, and must I REQUIRE Microsoft software?
Hell, no, and the day those "must haves" die out will be the day the market opens up and a broader base of end user inspired solutions will appear.

This is not a vague prediction.
Intel stumbled more than once (especially with their ridiculous proposition that 10 GHz P4s were possible), and Microsoft?
Hey XP wasn't bad, Vista...well it looked good (but how many of you like it), Win 7 with Aero was a good recovery (but I hope you all realize that its really just an advanced version of Vista).
Win 8, 8.1, 10 (with its undeletable electronic "assistant"), hey I really tried, and 10 is almost tolerable (although why Aero was dropped is anyone's guess), but frankly, all three of these are uglier than Vista, and not as functional as Win 7.

I want OUT.

I not buying a Windows variant that builds on this crap, especially one that tries to force me to make all my software purchase through an approved "store".

So, from here out its Linux, and Power 9 or ARM, so I can free myself from these shackles.

You all do what you want, personally I want to divorce myself from these lame standards and their perpetual promises that they'll "get it right in the next generation".

Feh.

Jim

[Iggy]

It shouldn't, that's exactly what version control is for. Version control isn't just using a version control tool, it's about doing so with a workflow. Usually you make your changes in your own branch, keep it up to date with any important upstream changes and when it's ready, it's merged back. Usually with some sort of review process that the project maintainers have made policy. Whether you are using a centralised repository model like svn or a distributed one like git, it's policy (and permission management) that prevent unreviewed code making it's way into production.

Thanks Karlos, I must admit that I'm old enough to remember when most projects were small enough to be handled by one programmer (or a lead programmer anyway).
This networked and cooperative environment shouldn't surprise me as its how all businesses operate these days.  Collaborative teams, electronic communication,centralized web sites...
I sir, am an apparent troglodyte in manner in which I prefer to function, which is still "Don't worry, I'll take care of it".
I have a profound distrust in the aptitude of others to get the job done and would prefer to handle as much of it myself as possible.

Its an attitude I've got to shake in order to function in this brave new world we've evolved into.

[Iggy]

Source control is very useful even with one person projects.

If you give each commit a decent description then it can remind you why you did something, without having to pollute the code with extraneous comments.

It can also be used as an infinite undo buffer, of course how much you can undo at a time is limited by how often you commit. It should be an end to those head scratching sessions while "this was working last week" goes over and over in your mind, or at least it will tell you exactly what changed.

Accidentally changing a line which doesn't stop it compiling/running but changes the behaviour are also much easier to detect.

Good points, it will be a part of anything I develop from this point out.
Beats digging around for earlier versions of your code (I'm not the most organized person).

[Iggy]

just fyi we are working on multiple chipsets so mpc107/tsi107 etc etc wont be the only supported

this will help with prices and availability of cards and we have someone working on cheaper pci carriers for those

Does  Tundra Semi still support and supply the mpc107/tsi107?

[Iggy]

thats iggy how we know him, right?

;)P

im curious if his aros ppc fork will share the fate of his appointed ppc custom hardware design. nevertheless, as i wished him luck, i hope it wont. looking at safari-leopard webikt diffs id really love to have some relevant big endian testing platforms. without too much bragging about. better doing.

Interesting you would reference that, as you all failed to respond in any significant numbers, and I ended up collaborating with some like minded Linux enthusiasts to  work on a  T2080 based laptop.

Funding for the first part of that, the scematics, is essentially complete and Acube has been working on the design of that.
I can thank the commitment of Roberto Innocenti and his friends for that, not Amiga enthusiasts because they always want someone else to shoulder the burden of creating the goods.

Linux enthusiasts don't think like that.

And I've  completed projects like this before decades ago with the 68000 and 68020 processors.

I have support from NXP that dates back to the days of Motorola Semi, and have full documentation of a couple of reference design boards (right down to the full board layouts) and the Cadence/Orcad software needed load and manipulate them.

So, what have you all been doing lately? Waiting for a project based on a third rate eMC500v2 based cpu?

I have, and will put my time and effort into directions that make sense.
That was part of a discussion via e-mail that started with Bill Buck over the weekend, about economics and sensible strategic business management.

BTW - That was my major, because there were no IT programs when I started in this field, Business Management.

And my first job offer was from Dr. Fred Hoffstetter, after doing some work/study while still in High School.
Fred eventually became the director of Multi-media Communications at the University of Delaware.

So, does anyone want to compare experiences, productivity, or the length of their penises again?

Thanks to coordination with Mark "Bigfoot" Olsen,  Paul "Acill" Rezendes, and David "AmigaDave" Morris we arranged a pre-release demo of MorphOS 3.10 at last year's AmiWest.

Again, I ask you all, what do you do with your spare time? Outside of work, I study for my MBA, and work on projects like these.

And I'm not particularly thin skinned, but if I set my mind to it, I usually get what I'm aiming for.

So, you all still waiting for third party salvation?
It ain't happening, we need to take it upon ourselves because  no one with a reasonable sense of economics and good business priorities would come anywhere near this community.

Btw, a 68K port of Oddesey is rediculous.

And to conclude, big median platforms are disappearing.
What next, you guys going to suggest we recreate those with FPGAs, only to still be stymied by issues related to software coded with little median platforms in mind?

Again, a move to a little median platforms is the only eventual course left to us with any sendibility.
Whether it's bi-endian like Power 9 or some ARM variants, or natively little median like X64, it doesn't matter to me.

I saw the writing on the wall in the '90's and while I didn't like it, I shifted my primary ISA.

My former employer died while still promoting 68K. Some of you seem dead set on that intent as well.

[Iggy]

They're defunct.  

The rest of the range can be found at IDT.

https://www.idt.com/products/memory-logic/ehb-embedded-host-bridges/?field-processor-bus=PowerPC%207xx%2F7448%20-167MHz%2CPowerPC%207xx%2F7448%20-200MHz&method-field-processor-bus=OR

You know, that was pretty stupid of me, because I have two samples of a 68K/PPC PCI bridge that used to be a Tundra product that were supplied by IDT.
I'm not sure where they are or if the vacuum packaging is intact, but that was one truly defunct idea I gave up on.
Too much code needed to drive the bridge.

And I apologize for the last rant.
Sometimes I have "flashbacks" to the '90s and that sense of immanent doom that everyone could see barreling down on my preferred ISA.
Sucks losing out to inferior technology because of "market forces".

But I'm not going to repeat that "last man on a sinking ship" experience.
It pretty much killed my friend and former employer, while I was in the 68K development community.
I tried to warn him (we'd received beta copies of Win 3.0 direct from IBM, that would load onto generic X86 hardware).
The birth of the computer as an appliance may have been born in the Apple Macintosh, but it was easy to see a future where to sole remaining major players would be the PC and the Mac.

Damned that sucked, and it Is still a gaping wound for me.

As to Bill Buck, some infer that he abandoned the community, but actually Genesi's devotion to the PPC nearly drove them into bankruptcy.

There are times when you have to realize it's over, and learn to move on.

Not sure I've completely learned that lesson yet, myself.