In the early '90s when my company was still fielding 68K based computers I might have agreed with you.
These days, the only individual I can think of who can pull off this kind of thing is Jens Schönfeld.

Great exsample!

M

Jim,
Quote:

Funny how one of the guys involved with the UltimatePPC managed to produce and sell a 1541-emulator for the C64, when Jens wouldn't... ;)

@stephen_robinson:

Quote:

...except when equipped with one of these babies.

The project is announced at too early stage to take it as a commercially viable one (even at hobby level). Well, it is for sure entertaining and educating for its developers. To say more, I guess they have a sincere intention to finish it. But, in my opinion, it will turn into a neverending story similar to NatAmi.

[ Edited by Krashan 28.05.2012 - 06:25 ]

> maybe they can change the PPC CPU to [...] G4's, or [...] G5 CPU's.

That would require a complete redesign of the board, which is further complicated by the fact that both (non-SoC) G4 and G5 need a northbridge chip to make them work. That aside, the QorIQ chip has been chosen due to its on-chip I/O capabilities (SATA, USB, Ethernet), effectively turning it into some kind of southbridge chip to the m68k chip. Neither G4 nor G5 can provide that.

Great point Andreas,
Many of Freescale's products built after the G4 offer this functionality.
Still not sure why they pick the SoC they intend to use when there are better products.

And I can't see how they'd make this work with the old libraries (like Warp3d).

> Still not sure why they pick the SoC they intend to use when there are better products.

Maybe those "better products" are more expensive, that may be a valid concern. My guess is that the UltimatePPC developers have just not been aware of the FPU compatibility issue between e500v2 and "standard" Power Architecture. Reading MorphZone would have helped them here as we have been discussing the issue since January 2009 ;-) One of the board's developers even confirmed that e500v2 had an FPU (which isn't a false statement per se, but "an FPU" may simply not be enough when it can't execute "standard" FP code):

http://translate.google.com/translate?sl=de&tl=en&u=http://www.a1k.org/forum/showpost.php?p=519941

> I can't see how they'd make this work with the old libraries (like Warp3d).

Warp3D wouldn't make sense anyway as the P1013's DIU has no 3D capabilities.


Edit: Btw, Troika once had plans for a standalone PPC board for the Amiga market based on e500v2 (PowerQUICC III, MPC85xx):

"In parallel with my work on Athena, I have been looking at cost effective higher performance system, codenamed Hercules. [...] The Hercules design is centred around an E500 core PowerPC running at a minimum of 1 GHz. [...] At this time the Hercules platform is in early development, i.e. schematics are in progress."
http://amigaworld.net/modules/newbb/viewtopic.php?topic_id=28102&forum=33&start=240#477216

And 2.5 years later:

"The Hercules design was little more than an early concept."
http://amigaworld.net/modules/newbb/viewtopic.php?topic_id=23127&forum=2&start=120#621485

[ Edited by Andreas_Wolf 30.05.2012 - 09:09 ]

It will be interesting to see what comes out of this.

P10xx and PPC460ex are affordable PPC chips, others are not.

But if 2x CPU price is not an issue then chips like MPC8610 and P2041 become alternative.

IIRC, some P10xx and P20xx are pin compatible. P2041 has e500mc core (compatible FPU), but not sure how easily it fits in place of 1013.

Any specs around of T1xxx chips yet? Do they have DIU?

amigadave,
Quote:
There are several reasons why it can not:
1) Both WarpOS and PowerUP are proprietary. Much of WOS is in powerpc assembly, making it harder to work with it even if they could obtain the source code
2) The FPU issue isn't specific to the OS or kernel being run. The CPU just doesn't implement the Power ISA category FPU. You cannot execute any regular PowerPC FPU code on it.

> P10xx and PPC460ex are affordable PPC chips, others are not.

Compared to the PPC460EX (PPC440 core), the newer Applied Micro chips with PPC464FP or PPC465 core are much cheaper.

> IIRC, some P10xx and P20xx are pin compatible.

Yes, P2010 and P2020 are pin compatible with P1011, P1012, P1013, P1020, P1021 and P1022. They all have e500v2 core.

http://www.freescale.com/files/32bit/doc/brochure/PWRARCHQIQSG.pdf

> P2041 has e500mc core (compatible FPU), but not sure how easily
> it fits in place of 1013.

P1013 and P2041 are not pin compatible.

> Any specs around of T1xxx chips yet?

I'm not aware of any.

> I don't even know if Woz was involved with the company when the
> G4 Macs were produced.

12 years passed between :-)

Andreas_Wolf,

Thank you for the pin compatibility info.

Untill someone does some Book-E spec FPU emulator for P10xx series.... or untill T1xx chips appear...

"newer Applied Micro chips with PPC464FP or PPC465 core are much cheaper."

Meaning APM86290 and APM86190? http://www.apm.com/products/embedded/multicore460/apm86290/
I once read very low price estimates on some early press info. (below 30 eur in high volume, IIRC)
But are those available yet?
I have not seen those being sold yet.
Perhaps you have a link to the price & availability info?


[ Edited by KimmoK 29.05.2012 - 16:13 ]

> Untill someone does some Book-E spec FPU emulator for P10xx series

Emulating standard Book-E/PowerISA FPU on e500v2 might be hard (or impossible even?) as this is not simply about missing instructions here but about instruction overlappings, i.e. same instructions doing different things on different cores. This seems similar to m68k vs. ColdFire. For an exhaustive and detailed comparison of the programming models of e500v2 vs. e600, refer to:

http://www.freescale.com/files/32bit/doc/app_note/AN3531.pdf

Btw, even the (ex-)maintainer of Debian for PowerPCSPE calls it a "braindead FPU design":

http://lists.debian.org/debian-powerpc/2012/05/msg00025.html

>> newer Applied Micro chips with PPC464FP or PPC465 core are much cheaper

> Meaning APM86290 and APM86190?

Yes, my statement was mainly about the Mamba chips (PPC465 core), but also APM82181 (PPC464FP core, although with only one single PCIe x1 interface and no PCI that SoC would probably not be satisfying).

> I once read very low price estimates on some early press info.
> (below 30 eur in high volume, IIRC)

I guess you might be confusing this with the APM821xx announcement from some months before:

"Limited sample quantities of the APM821xx family are available at less than $20 for quantities of 10,000 pieces."
http://investor.apm.com/phoenix.zhtml?c=78121&p=irol-newsArticle&ID=1414268

AFAIR, nothing of that kind was ever announced for the APM86x90 chips.

> are those available yet? I have not seen those being sold yet.
> Perhaps you have a link to the price & availability info?

"Avnet is quoting $77 for the single-core APM86190-1200 and $120 for the dual-core APM86290-1200 for qty 120."
http://amigaworld.net/modules/newbb/viewtopic.php?topic_id=31667&forum=33&start=260#619403

I just had a look at Avnet and saw that APM86x90 chips (still only available up to 1.2 GHz) became cheaper since that statement quoted above was made:

1.2 GHz APM86190 at qty of 120: 58...64 USD
1.2 GHz APM86290 at qty of 120: 87...96 USD

Piru,
Quote:

So my uninformed assumption that this board is useless appears to be true.

Either the two guys working on it are unaware of how futile their work on this project is, or this whole thing is just a joke they are playing for some strange reason.

Given the fact that at least one of the two has completed successful projects in the past, the lack of knowledge that it can't work is puzzling and the assumption that it might be just a joke does not make any sense either.

Oh well? Someone will figure it out eventually and I will read what is really happening somewhere.

Unless someone designs a new Cyberstorm accelerator with a 1.5GHz G4, or faster CPU, plus 1.5gb RAM, USB2.0 and a SATA controller built-in, I don't see the point in creating any thing less than that for the A3000/A4000 Classic Amiga models.

I have a feeling being HW developers they didnt consider SW side carefully... PPC could run UAE under Linux or perhaps AROS so you could run 68k software emulated on this card but then there is question why have that 68040 there at all.

Perhaps APM82181 would be the best option for UlitimatePPCaccelerator kind of HW and for some PPC-efika killer ultra-low-end below 200 eur HW. (perhaps there also exist some suitable mobile GPU for miniPCIe connector, inside bigbox classic, or ePCIe should be used on classic accelerator kind of HW) http://www.hwtools.net/jpg/PE4H-PM3N-ver2.4.jpg http://i.imgur.com/zFKr3.png (3d bench, red is for separate PCIe card and blue is laptop internal, more info http://i.imgur.com/4B7hU.png )

And APM86190/290 start to look like the best chip for low...mid range HW around 200...300eur.

Too bad that for now there is no good options for single board computer designs (Rasberry like). Unless APM86x90 is built with some mobile Radeon onboard.Finnish page for the rest.

[ Edited by KimmoK 30.05.2012 - 09:07 ]

KimmoK,

For Rasberry pi alike designs there's still the 5125. It's not classes beyond and cheap enough. But there are actually two major show stoppers: no hw video decoding unit and no cache coherence (and this is at least a fail for MorphOS). But apart from that a mini board for really little money would be pretty possible. About a year ago I was really considering starting such a project, I spend quite some hours reading a lot and calculating this and that, but eventually I decided against that project for several reasons.

> Perhaps APM82181 would be the best option [...] for some PPC-efika
> killer ultra-low-end below 200 eur HW. (perhaps there also exist
> some suitable mobile GPU for miniPCIe connector [...])

As you'd need audio as well you'd have to attach some combined (2D) graphics/audio solution to the PCIe x1 interface, similar to the SM502 on the Sam460ex ...or use the APM82181's SPI interface with a CPLD acting as SPI-to-I²S bridge and connect an audio chip via this I²S ;-)

About audio: yes, solving the audio problem without loosing PCIe of the 3D expansion is one issue. Nice idea about that SPI. Perhaps audio via USB could be another or Audio via localbus connected xcore.

"spend quite some hours reading a lot and calculating this and that"

I think I'm going through same hobby engineering/research fun ...

For me, I would like the HW to be a little faster than my curent SAM440@667Mhz.
(goals 1:learn HW 2:do some designing 3:try to make/get a prototype & OS running on it 4:see if I could manage to find means for actual production run 5:provide affordable opensource PPC HW 6:keep it PPC AmigalikeOS compatible, just in case if one day....)

I'm going to be buying one of these, it is perfect for my needs.

I still enjoy using my 3000 for some functions even though I can do that same on my Peg II. It will be fun to put some new hardware in it.

For me:
- 33 MHz 040 is plenty fast for what I use the classic hardware for.
- 3000's are very picky about accelerators and this looks like it will fit and give modern I/O and fast memory.
- 3000's with a failed 030 can use this to bring it back to life.

I would think that these guys who are developing these, understand this target market potential. Great idea for them to come up with and hope they sell tons of them...

Even if the PPC chip is incompatible with others, PPC part could be usable on opensource SW use...
If they manage build a dongle to replace A3k/4k motherboard, they might have some extra market from PPC Linux / PPC AROS fans.

(the dongle could come with NATAMI/miminmig FPGA....)

> If they manage build a dongle to replace A3k/4k motherboard, they
> might have some extra market from PPC Linux / PPC AROS fans.

But then, PPC Linux or PPC AROS users won't need the m68k part of the board which only makes the board unnecessarily expensive for them. If PPC Linux or PPC AROS users for whatever reason really want to run their OS on QorIQ P1013/P1022 (or any other e500v2-cored chip) there are already numerous solutions out there.

matt3,
Quote:

I'm glad you're such an optimist.
Personally, i think this is an elaborate hoax.

BTW - What do you intend to do with the PPC (since its incompatible with existing software)?

>> This thing is completely useless if all it can run is WarpOS software or
>> PowerUP software. Can it even run those?

> There are several reasons why it can not:
> 1) Both WarpOS and PowerUP are proprietary. Much of WOS is in powerpc assembly,
> making it harder to work with it even if they could obtain the source code
> 2) The FPU issue isn't specific to the OS or kernel being run. The CPU just doesn't
> implement the Power ISA category FPU. You cannot execute any regular
> PowerPC FPU code on it.

From the current status update:

"Both delevopers [...] brought along Samo Jordan, original developer of the WarpOS/WarpUP! software. Current status is:
[...]
- Active talks with Hyperion for supporting AmigaOS4.x.
[...]
- Support for WarpUp/WarpOS through support pledged by the original author"
http://www.amiga.org/forums/showpost.php?p=696776
http://www.a1k.org/forum/showpost.php?p=528160