Jim wrote:,
Quote:


I wonder why the people responsible for making the design decisions for the final X1000 motherboard design ignored Paul's advice to stay away from the PA6T and look at using one of the QorIQ chips instead? Perhaps the only reason was because using one of the QorIQ chips would have meant giving up PCIe lanes and also meant further delays in completing the X1000 project.

What ever the reasons for staying with the PA6T, I am glad that I ordered and received my X1000 when I did, as there will most likely not be any more of them produced, and I don't know how long it will take A-Eon to complete the design, production of beta boards, and the beta testing of their next system(s), which are planned as replacement systems for the X1000 design.

I favor Andreas' belief that they had they processor in mind before they contacted Paul.
And Varisys, as an existing customer, could order them.
Wasn't this one of the chips mentioned by Amiga Inc. in relation to Ack systems?

[ Edited by Jim 01.03.2013 - 17:13 ]

> I wonder why the people responsible for making the design decisions for
> the final X1000 motherboard design ignored Paul's advice to stay away from
> the PA6T and look at using one of the QorIQ chips instead?

It's easy to understand if you look a the timeline:

- Early 2008: A-Eon's decision for the PA6T is made by the OS4 developers (as requested by A-Eon)
- Early 2009: actual development of Nemo board commences at Varisys
- April 2010: Paul Gentle of Varisys discourages from using the PA6T for new designs
- Mid-2010: Paul Gentle of Varisys recommends QorIQ

If you then add to this that the first somewhat suitable QorIQ CPU (P4080) started sampling in August 2009, your question should be answered in full.

You can find more information (e.g. on the viable CPU options besides PA6T back then), links and Trevor's confirmation there:
https://morph.zone/modules/newbb_plus/viewtopic.php?forum=3&topic_id=7348&start=85

> Perhaps the only reason was because using one of the QorIQ chips would have meant
> giving up PCIe lanes

While the matter of PCIe lane quantity is a valid one, I believe the reasons stated above outweigh it by far.

> and also meant further delays in completing the X1000 project.

At least delays in commencing, as with QorIQ instead of PA6T, the actual board development would have started at least 6 to 12 months later than it had with the PA6T. And this with a slower CPU (per core), no AltiVec and less PCIe lanes. Whether QorIQ would have meant less problems during the design process, I don't know.

> I favor Andreas' belief that they had they processor in mind before they contacted Paul.

At least that's the story as told by Trevor. He said that PA Semi (before they were bought by Apple) had referred A-Eon to Varisys as a design company familiar with the PA6T.

> And Varisys, as an existing customer, could order them.

Yes. While A-Eon could have ordered from PA Semi, after PA Semi was purchased by Apple only pre-existing customers were allowed to order.

> wasn't this one of the chips mentioned by Amiga Inc. in relation to Ack systems?

Yes, it was.

https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=6775&forum=11&start=10

Hard to believe that Trevor actually made a Bill McEwen pipe dream come true.
No matter what you think of the value of the X1000, the fact that he succeeded in getting it to market is impressive.

When Andreas first started relaying some of these rumors to me I remember thinking "No way, I've exchanged messages with Paul before, and while he has designed PA6T based systems, he has been recommending using other processors".

One thing I learned over time, don't discount something Andreas puts forth. He put a lot of research into hid statements and they are rarely based solely on his opinions.

Usually, he's a few months ahead of everyone else on the information learning curve.

And his indexing ability has made him invaluable in double checking the veracity of a point or position.

Derringer,
Quote:

I started with much more tedious tasks, like matching the pinouts between ATI and Uli Southbridge components (and noting required glue circuit changes) and making changes to Freescale schematics.
What I've got primarily are a ridiculous amount of written notes and few schematics (which after several months work still are not complete), and they beginning entries into some layout software.

And, it would have been much easier if I had access to some of the layout and design packages Freescale had used. It didn't occur to me to make an inquiry till a couple months into this, but I think it would have been possible to negotiate access to Freescale's layous.

In a way, Freescale is very much like Motorola.
They offer assistance and support with a lot less complication and qualification then their competitors.

> Usually, he's a few months ahead of everyone else on the information learning curve.

I'm still somewhat ashamed about me having been so much behind the curve regarding the e5500 core (aka "e500mc64"). The web was full of hints yet I somehow managed to miss them for months :-)

>The web was full of hints yet I somehow managed to miss them for months

You were still ahead of me on this one, and I had already been pointed toward the Qorlq line earlier.
So far, e5500 and e6500 cores have been the best PPC products released in current years.
I'm getting tired of perpetually being disappointed by Applied Micro.

I wonder if Freescale has considered a mixed PPC/ARM product?

amigadave,
Quote:

Yes, I think people tend to overlook that when considering the X1000.
Its got two PCIe x16 connections.
Very cool and unlikely to be matched by future successors.

> I'm getting tired of perpetually being disappointed by Applied Micro.

They have long said goodbye to Power Architecture in terms of new developments. For them it's internally mostly been about ARMv8 for some years.

> I wonder if Freescale has considered a mixed PPC/ARM product?

The announced QorIQ Layerscape family may provide pin-compatibility between PPC and ARM members but I doubt that chips with PPC and ARM main cores on the same die will come. What's already there for some years are chips with PPC main core(s) and ARM security core from Applied Micro.

> the X1000 [...] got two PCIe x16 connections.

Connectors, not connections. Connections are either one x16 (with the second slot disabled) or two x8.

> unlikely to be matched by future successors.

To match the x16 speed of the PA6T's 1st-gen PCIe it only takes x8 speed of QorIQ P and T1's 2nd-gen PCIe and even only x4 speed of QorIQ T2+'s 3rd-gen PCIe.

Good Point about speed Andreas.
But even a first gen x8 PCIe connection is adequate for most cards.

Heck, I'm even seriously considering a SAM460 with a Radeon X1300 (X4 first gen + slow video card).

Btw - When do we get USB 3.0 support?

> even a first gen x8 PCIe connection is adequate for most cards.

That's same speed as x4 on QorIQ P and T1 or x2 on QorIQ T2+. Should be doable.

> I'm even seriously considering a SAM460 with a Radeon X1300

Let's wait for the R500 drivers first ;-)

>Let's wait for the R500 drivers first

OS4 2D drivers already exist.
But then, you know that not my primary interest. ;)

>> Let's wait for the R500 drivers first ;-)

> OS4 2D drivers already exist. But then, you know that not my primary interest. ;)

Exactly. That's why I was thinking you were alluding to MorphOS on the Sam460ex rather than to OS4 :-)

> That's why I was thinking you were alluding to MorphOS on the Sam460ex rather than to OS4

Well, let us hope that they will not make us use PCI video cards with that board.

Frank did tease us once mentioning X600 drivers in regard to (I think) iMac.

I'm just an optimist.

X600 would probably be easy.
There are R400 cards that work in that slot (so a driver could work on both G5 Macs and the SAM460), but these are not that cheap.
A cheap R500 card like the X1300 would cost no more then an X600 and perform better.
Also, most are passively cooled (as is the rest of the SAM460).

But I'd be willing to buy whatever is supported by both OS'.

We will have to wait to see what graphics cards hey support for this system. ;)

> Frank did tease us once mentioning X600 drivers in regard to (I think) iMac.

Not only mentioning but even showing screenshots of such driver in action:

https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=9034&forum=11&start=6