Sonnet crescendo 7200 and OS4.1

[Iggy]

So, I am unclear on this.
Will Hyperion eventually support the use of the Sonnet crescendo 7200 when installed on PCI enhanced legacy Amigas?

Sonnet cards range as high as 800 MHz using G4 processors.

Were these boards to be reverse engineered, there are still compatible low wattage G4 processor available that run at 800 MHz and 1 GHz that could be utilized.

This option would bring the cpu speeds up to levels comparable with lower end AmigaOnes.

With full legacy chipset support, PCI expansion, and high speed PPC processors, legacy hardware would be able to run far more Amiga software than any NG platform, at speeds that would make running everything from OS3.1 to OS4.1 practical.

[Iggy]

So the only utility is in legacy applications using Warp3D?
That's a lot of expense and effort to run a very small number of programs.

That knocks the Sonnet out of consideration for me.

In order for a PPC to access the chipset, I'd assume it needs to be incorporated into the processor card, right?

Reverse engineering the Crescendo is a whole lot less attractive to me at this point.

[Iggy]

The first consideration for the Sonnet Crescendo was long long before the Shark.
Let's just say that there were disagreements on how to proceed and leave it at that.

Years later inquiries were sent out as to existing stock and condition of same.
Non-starter. There are people here who can supply the exact figures.

#6

The low wattage version of the cpu used (in 800 and 1000MHz speeds) is available in adequate quantities.

[Iggy]

Surely you mean WarpOS?

Actually, I was thinking in terms of MorphOS Goa, but yes I should have referred to WarpOS.

Considering there is basically ZERO availability of Sonnet's why would someone put time/money into this type of development?

I would not mind it in my tower but...

Get yourself... A1222 when (if) it will be out.

Tabor? Obviously you're not familiar with my posts.

I will NEVER own one of those boards.
I'd consider spend the money on an X5000, but a system based on the crippled P1022?
Why would I want a system with a 32 bit cpu and a non-stand fpu, when the last two AmigaOne systems have had 64 bit cpus with standard fpus?

Because it costs less? So do lime green polyester pants from Kmart, hey they might give you two for the cost of one (item that is much less ugly).

But if cost was a primary issue, I'd stick with the Raspberry Pi 3.

But back to the topic, if legacy PPC boards can access the chipset directly and the Sonnet Crescendo can't, then I have no real use for the Sonnet board.

It doesn't appear that we have the talent left in the Amiga community needed to build a new 68K/PPC based processor slot board.
So, my last PPC system may be an X5000 (hopefully an X5000/40 if it gets produced).

Then I'll follow the MorphOS migration to X64, while keep my peripheral vision trained on what hardware comes after Tabor.

I'm actually pretty fond of what we've managed with PPCs, but its probably time to move on.

[Iggy]

The Sonnet 'team' are actively working on getting faster G4 PMC cards running, and there is even talk of them producing a bespoke very fast (1Ghz+) card at some point.

Well, there are still completely compatible low wattage 1GHz cpus available from NXP.
And reverse engineering the board?
Remove the components and send it to the proper company.
It's doable.

But a board with no direct access to the chipset, where everything native has to be directed by the 68K?

That doesn't seem nearly as useful as a processor slot board.

[Iggy]

Thanks for the multiplicity of clarifications gentlemen.

[Iggy]

It seems like an OS4.1 port is completely out of the question. It would be cool though if these cards could be harnessed as coprocessors under OS4.1 Classic, the same way as they are currently under OS3.9 to run specific WarpOS programs.

Interesting thought. It would be more powerful than a standard 603/604 PPC processor. Without calling it SMP, could two PPC cpus be implemented under OS4? The cpu plugged into the cou slot would obviously have to be the master, but I can't see why some tasks couldn't be farmed out to the G4.

[Iggy]

How many customers can possibly use this? 1 or 2? Have you ever seen the correct Sonnet card for sale? I've found a couple mislabelled as this on ebay but were incorrect. Does it make sense to support such a board that effectively no one can run OS4 or even OS3.1 on?

I've seen a few, but the majority needed to be repaired.
Currently I don't own an Amiga with a PCI expansion bus.

But I'd assume that anyone with a 4000 with a mediator would be a possible candidate, so I'd assume its more than "1 or 2".

Exploring using the card isn't something we can vote on, the team working on it wants to use it, so its getting done.

My basic question was could it be used for something more than WarpOS packages, since there are only a few of those.

Personally, if a new PPC board for legacy Amigas was created, I'd rather have something that fits in the processor slot and has access to the chipset.

An A4000 equipped with a 1GHz G4 would be very competitive with some AmigaOne's and it would be able to run software that hits hardware directly.
Add a mediator and an RTG card and you've got a legacy system that is as powerful as an AmigaOne, but retains register level compatibility.

[Iggy]

The fast access to Voodoo graphics memory would be lost with an A4000 accelerator slot unless someone also built a GREX like replacement PCI busboard.

Fast? Funny, I never thought of 33 MHz as fast, but I see your point, faster than accessing the PCI bus through some bridge.
And it would be nice to have support for something faster than a Voodoo3.
Would a Voodoo 4 4500 work?

[Iggy]

At which point (PPC-CPU+PCI+access2AmigaMobo) you'd be back at what the Escena-A1 was supposed to be.


Well that was a pointless brainfart 20 years ago........

Can you point me to technical info on that?
It doesn't sound like they had the resources to realize it, but the idea sounds practical.

I wasn't really paying attention to the Amiga community at that point.
All I've heard about that project merely made it sound like a scam, but a "pointless brainfart"?
What seems pointless, is that when you tack a PCI bus onto an Amiga everything is coordinated via a bridge to the much slower native bus.
Why not have a direct PCI connection to the PPC cpu?

[Iggy]

This article translate from French gives a good overview over the project and how things panned out.

http://translate.google.fr/translate?u=http://obligement.free.fr/articles/amigaoneppc1200.php&sl=fr&tl=en&hl=fr&ie=UTF-8

There are some extra photos on the bottom of this page.

https://translate.google.co.uk/translate?sl=auto&tl=en&js=y&prev=_t&hl=en&ie=UTF-8&u=http%3A%2F%2Famiga-mania.orgfree.com%2FNoticias%2FAmigaOne%2FAmigaOne.html&edit-text=&act=url

You may have already seen the schematic before on the Phake5 Facebook page.

Interesting, but now I understand Kronos' objections. There is virtually no integration between the legacy system and the PPC side of the system.

At that point, the stand alone AmigaOne or Pegasos makes as much sense.

It almost would have made more sense, if the two were to be integrated, to created a single card legacy system that could be plugged into one of those PCI sockets.

Now, maybe we could examine that comment about running Crescendo cards in an X86/X64 platform. If PPC specific software could be run on a card, and the rest of the OS ported to the X86 ISA we'd have an platform that would allow the use of cheaper, more powerful commodity hardware for newer applications, and still have compatibility with older applications without the necessary complication of emulation.

Better still, one plug in board similar to the Vampire for OS3.X compatibility.
A PPC card for OS4.X or MorphOS compatibility.
All running in an X64 system that could handle the next generation of MorphOS or maybe AmigaOS 5?

[Iggy]

An A1/Pegasos makes "much sense" the moment you replaced all things Amiga with much more powerful components from 15 years later making the Axx00 an glorified keyboard controller.
(same reason I don't get all the buzz about the Vampire)


Now Phase5PPC+Permedia/PCI does already fall into that category but is acceptable due to being released in a different time.


Bout having a Sonnet-PPC in an MorphOS-NG running x86-64, just start MorphOS3.x in QEMU and be done with it.

I kind of get Vampire, until you reach Vampire4, where attachment to legacy hardware is redundant as all functionality now resides in the FPGA and it can exist as a stand alone system.

And yes, QEMU is beginning to look like the solution for running current NG software.
Although the first project that intrigues me is the idea of emulating the SAM460 and the X5000 on a PCIe PowerMac G5 11,2.
And I'd assume similar emulations could been done easily on a Power 9 based platform, possible multiple instances of OS', one or more than one type.

How about ALL Amigoid OS running on one platform with plenty of threads to spare.
Since Power 9 has PPC compatibility, our current NG OS' and hardware (outside of AROS) would be easier to emulate and require less cpu power than similar  emulation via X64.

We would need something that would run above these emulations, provide a common user interface and a method of compositing the various display windows, and accessing the hypervisor.
Possibly a variant of Linux.
OR Bigfoot, Keiro and a few other talented programmers could expand on Ambient and and an SMP enabled kernel could be provide the core of something beyond MorphOS that manages everything, and provides an environment for 64 bit code with memory protection and all the modern features we are currently missing.

Nothing like that is likely to happen, as we are all mired in this slow death march of evolution that is one step from completely frozen, dragging around the corpse of a legacy API.

But you're right, one system would do.

I guess I'll start exploring the options right now on my X64 hardware and my 11,2 PowerMac.
If the Amiga community can't get its act together on hardware platforms, the rapid advance of platforms outside our community should provide enough power to emulate anything we run within our various camps.

So guys, I'm still focused on that X5000/40. After all, that will run MorphOS, OS4, and Linux.

But, once you all embrace low end hardware like the A1222 (which will only run OS4 and Linux, and then only moderately well), I will hopefully have better hardware, and will be working on a unified emulation of everything you have with a platform that can run that and more.

Good luck with the direction so many of you want to take.
The goal is to pedestrian for me.

[Iggy]

Hey, 2005 called and wants it's thread back.

Thanks Karlos, personally I wanted PS3 support, until I found out how little ram it had. And the Xbox360 also looked interesting, but documentation for the hardware of either wasn't great.
WiiU? Hmm, slower, unexpandable, and my G4 PowerMac is probably faster.
What would be the point?

@ Kronos, X64 Qemu emulation of PPC hardware looks like a winner, and at least we aren't targeting a dying ISA.

I like PPCs, and Power 9 would have been a cool linear upgrade, but following the MorphOS development team's lead on what hardware should receive support has brought me affordable, function hardware from mainstream computing companies.

So I guess I'll stop worrying about brand A, and just stay the course.

Of course, if QEMU emulation of the X5000 becomes a reality on the 11,2 PowerMac, I have a Quad 2.5 GHz unit sitting here waiting.

[Iggy]

Now you are talking!

That would be ironic, wouldn't it? Emulating a $1700 2018 PPC system with a 2005 system that can be had cheaper and is more powerful.

[Iggy]

Are you talking to me or the OP?   HA!

20 Million WiiU's running a triple-core PPC @ 1.24 Ghz for about $150 or you can spend hundreds of dollars on a card that's not much more powerful than a Gamecube...  Makes sense.

By the way, there isn't much in the way of porting needing to be done.  Mostly just UBOOT.
https://en.wikipedia.org/wiki/Das_U-Boot
...much like MONA very little work would be required.  "Drivers" amount to calling the libraries that already exist within the machine's ROM.

I know Amigans aren't used to getting alot of feature for little money and when you do, you make sure that you complain heavily (ie.. Vampire2)...

You know what I'm wrong.  I don't want:
2GB of RAM
Out-of-order execution PowerPC based cores
45 nanometer process technology
IBM silicon on insulator (SOI) technology
Three cores at 1.243125 GHz
Symmetric multiprocessing with MESI/MERSI support
Each core can output up to 4 instructions per clock using superscalar parallelism.
32-bit integer unit
64-bit floating-point (or 2× 32-bit SIMD, often found under the denomination "paired singles")
A total of 3 MB of Level 2 cache in an unusual configuration.[16]
Core 0: 512 KB, core 1: 2 MB, core 2: 512 KB
4 stage pipeline
7 stage pipeline - FP
6 Execution Units per core (18 EUs total)
Die size: 4.74 mm × 5.85 mm = 27.73 mm^2
4 USB ports, 802.11N wifi, bluetooth... dual-layer BluRay drive...
...and I surely don't want a 550Mhz core Radeon Latte GPU...

I don't want any of that for less than $5000!!!!
How dare that be available for <$200!!!  What a travesty!

Still looks dated.
Only a 1.24 GHz cpu that does not appear to be significantly more powerful than NXP's e600 and e500 cored processors (although the L2 cache is interesting), and no NG OS supports SMP, so the core count is irrelevant.
I have been unable to research the fpu, to see if it similar to other PPC floating point units, or if it has AltiVec instructions.
The gpu, while heavily customized, resembles an AMD R700 and from my initial examination appears to only have 32GB of vram.
New video drivers would have to be created for this, as there are significant differences from a standard Radeon gpu.
Customized drivers for sound, networking, and possibly USB would be necessary.
The Blu Ray drive is pointless, as there are no NG drivers for Blu Ray, only DVD.
Oh, and you forgot to mention, its got no hard drive.

It has no expansion slots.
Basically, right now I have a dual cpu G4 PowerMac running at 1.33 GHz, with a VIA Envy24HT sound card, SATA SSD drives, SATA DVD-R drive, a wireless networking card and an R500 video card.
The basic system with a MorphOS license cost me less than $100. And it all works with existing drivers.

If I want to spend something closer to the $200 you're mentioning, I can buy a 2.3 to 2.7 GHz G5 PowerMac, that can take up to 8 GB of memory, comes with interfaces for PATA and SATA, can handle the same R500 based video card, and may soon be able to use an AGP R600 video card.

So no, I don't want to buy a Nintendo game machine. Unless its to play video games on. I already have computers.

And I don't want to see the OS developers I rely on wasting their time porting to said under-powered game machine when there is so much other work that could be done.

So yes...no.