X1000
  • MorphOS Developer
    itix
    Posts: 1516 from 2003/2/24
    From: Finland
    Quote:


    I take it you refer to the OS level here, right? The reason I find Dickinson's statement odd is that it seems to refer to the hardware level, not to the OS level. If A-Eon really have fully functional dual-core PA6T chips (which I believe to be the case) then there shouldn't be any "tricks" needed to get both cores working properly on a hardware level.



    Indeed and I am aware of rumours... I just have my doubts elsewhere.

    Quote:


    First, we should define what this "game" is about at all and what "winning" should mean in this context ;-) I'd say that if A-Eon manage to release even a single batch of X1000 machines to real end-users (as opposed to beta testers) then this could be called a "win" for them to some degree.



    That is a win, sort of. Many start up companies never get out of so called Death Valley. Some companies never get even that far. Nevertheless Amiga companies keep dying in Death Valley no matter what.



    [ Edited by itix on 2011/3/3 7:40 ]
    1 + 1 = 3 with very large values of 1
  • »03.03.11 - 04:40
    Profile
  • MorphOS Developer
    Krashan
    Posts: 1107 from 2003/6/11
    From: Białystok...
    BTW Varisys website does not list any PASemi based products anymore.
  • »03.03.11 - 05:34
    Profile Visit Website
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12058 from 2003/5/22
    From: Germany
    > Varisys website does not list any PASemi based products anymore.

    Well spotted. Apparently, they did a complete website overhaul 3 weeks ago. The removal of their PA6T based VM31 board is in line with what they had told to Jim, i.e. that A-Eon had to source its own PA6T chips.
  • »03.03.11 - 10:12
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12058 from 2003/5/22
    From: Germany
    Addendum:

    > what they had told to Jim, i.e. that A-Eon had to source its own PA6T chips.

    Mathias 'Corto' Parnaudeau seems to object to that:

    "l'AmigaOne X1000, conçu par la société Varisys, ex-client de PA Semi et détenteur d'un stock de processeurs à ressusciter."
    http://obligement.free.fr/articles/pa6t-1682m.php

    Google translation:

    "the Amiga X1000, designed by the company Varisys, former PA Semi client and holder of a stock of processors to be resurrected."
    http://translate.google.com/translate?sl=fr&tl=en&u=http%3A%2F%2Fobligement.free.fr%2Farticles%2Fpa6t-1682m.php
  • »07.03.11 - 14:24
    Profile
  • Order of the Butterfly
    Order of the Butterfly
    discreetfx
    Posts: 388 from 2003/7/26
    From: Chicago, IL
    I heard the MorphOS port to the X1000 was further along than the OS 4 port.


    :)
    DiscreetFX
    Making your
    Digital Films
    More Effective!
  • »08.03.11 - 02:53
    Profile Visit Website
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4977 from 2009/1/28
    From: Delaware, USA
    Quote:


    Andreas_Wolf wrote:
    Addendum:

    > what they had told to Jim, i.e. that A-Eon had to source its own PA6T chips.

    Mathias 'Corto' Parnaudeau seems to object to that:

    "l'AmigaOne X1000, conçu par la société Varisys, ex-client de PA Semi et détenteur d'un stock de processeurs à ressusciter."
    http://obligement.free.fr/articles/pa6t-1682m.php

    Google translation:

    "the Amiga X1000, designed by the company Varisys, former PA Semi client and holder of a stock of processors to be resurrected."
    http://translate.google.com/translate?sl=fr&tl=en&u=http%3A%2F%2Fobligement.free.fr%2Farticles%2Fpa6t-1682m.php


    That article also claims that the PA6T does not need a Southbridge while the X1000 has an AMD SB600 onboard.
    Its a well written article, but I still find Varisys to be an unlikely source for PA6Ts.
    "Never attribute to malice what can more readily explained by incompetence"
  • »08.03.11 - 03:06
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4977 from 2009/1/28
    From: Delaware, USA
    Although Varisys' website does have this statement:

    "Intelligent "design led" procurement
    Backed by a team of experienced design engineers our procurement department can ensure that parts are available to meet you build schedule no matter how demanding. Managing the supply chain is just the start, suggesting design changes to cost engineer products or alternative parts where possible all count towards our value add procurement service."
    "Never attribute to malice what can more readily explained by incompetence"
  • »08.03.11 - 03:15
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12058 from 2003/5/22
    From: Germany
    > That article also claims that the PA6T does not need a Southbridge
    > while the X1000 has an AMD SB600 onboard.

    Yes, strange. And another odd claim:

    "PA Semi n'ayant pris qu'une licence d'architecture auprès d'IBM"

    Google translation:

    "PA Semi did not take an architectural license from IBM"

    Actually, P.A.Semi did take an architectural license from IBM. What they didn't take from IBM is a core license (which is what Applied Micro, LSI and others took).


    Edit: Google translation is wrong, see https://morph.zone/modules/newbb_plus/viewtopic.php?forum=3&topic_id=7183&start=212

    [ Edited by Andreas_Wolf 03.05.2011 - 23:07 ]
  • »08.03.11 - 09:32
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4977 from 2009/1/28
    From: Delaware, USA
    I missed that, but that's obviously wrong as they'd have to have a license to cover a Power based design.

    I don't know if we are ever likely to find out how A-eon sourced these CPUs, but if it was through Varisys then they may have trouble obtaining them in the future.

    The X1000 looks like a dead end even before its shipped.
    "Never attribute to malice what can more readily explained by incompetence"
  • »08.03.11 - 15:56
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12058 from 2003/5/22
    From: Germany
    > if it was through Varisys then they may have
    > trouble obtaining them in the future.

    Yes, and moreover that'd mean they served up some fairy tales (NDA regarding CPU, CPU supplier suggested Varisys as design company).
  • »08.03.11 - 17:43
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4977 from 2009/1/28
    From: Delaware, USA
    Yes, no matter what the actual series of events the statements contradict each other so something is plainly untrue.

    It makes me uncomfortable as I had high hopes for this machine. It's curious that if it had been designed solely by Varisys it would have like featured a Freescale QorIQ processor (since that was their recommendation when I was in contact with them).

    Trevor may have made a serious error using the PA6T.
    "Never attribute to malice what can more readily explained by incompetence"
  • »08.03.11 - 17:53
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12058 from 2003/5/22
    From: Germany
    > It's curious that if it had been designed solely by Varisys it would have like
    > featured a Freescale QorIQ processor (since that was their recommendation
    > when I was in contact with them).

    I'm afraid you're ignoring the chronology here. Varisys' recommendation of QorIQ to you took place 1.5 years after they started to design the Nemo board, according to A-Eon. 2 years ago there simply was no suitable QorIQ silicon available.

    QorIQ P1 and P2 (not suited for our purposes):

    "The first members of the P2 series are expected to sample to lead customers in Q4 2008, with beta samples scheduled for Q1 2009. The first members of the P1 series are expected to begin sampling in 2009."
    http://media.freescale.com/phoenix.zhtml?c=196520&p=irol-newsArticle&ID=1165851

    QorIQ P4:

    "The QorIQ P4080 processor is scheduled to begin sampling in mid-2009."
    http://media.freescale.com/phoenix.zhtml?c=196520&p=irol-newsArticle&ID=1165848

    "Aug. 27, 2009 – Freescale Semiconductor is now sampling its QorIQ P4080 multicore processor"
    http://media.freescale.com/phoenix.zhtml?c=196520&p=irol-newsArticle&ID=1324841

    > Trevor may have made a serious error using the PA6T.

    I can't really criticize him for having decided for that CPU about 3 years ago. I think that decision was made even before Apple bought P.A.Semi. Do you think he should better have switched to another available Power Architecture CPU for the X1000 prior to commissioning Varisys two years ago, like MPC86xx or PPC970? If yes, do you really think the X1000 project would be more advanced by now?
  • »08.03.11 - 18:59
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4977 from 2009/1/28
    From: Delaware, USA
    All very valid points. I don't think the average hobbyist has any idea how long it takes to produce such a complex product with limited resources on a small scale.
    And when you consider that PA Semi's target for the PA6T was Apple and Apple surprised them by moving to X86, it didn't look good for the company's long term prospects even back when Trevor may have first considered this processor.

    Apple has not committed to keeping this product available over a long term. Nor have they considered licensing it out. The PA6T looks dead.
    "Never attribute to malice what can more readily explained by incompetence"
  • »08.03.11 - 20:42
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12058 from 2003/5/22
    From: Germany
    > when you consider that PA Semi's target for the PA6T was
    > Apple and Apple surprised them by moving to X86, it didn't
    > look good for the company's long term prospects even back
    > when Trevor may have first considered this processor.

    Actually, most people thought it looked good for them. Mind you, the PWRficient was only introduced about half a year *after* Apple had announced their switch to Intel, so the public didn't take it for a new chip for Apple anyway, but all along for a chip for embedded purposes. And that's exactly what the PWRficient was introduced as by P.A.Semi (the original purpose of the chip was only revealed at a later date).
    According to A-Eon, the decision for the PA6T was made about 2.5 years after the chip was introduced, which was about the time it went from sample state to production state.
    Overall, I don't share your assessment that "it didn't look good for the company's long term prospects".

    > Apple has not committed to keeping this product
    > available over a long term.

    It was reported they were obliged to keep it available at least until mid-2011, and until mid-2013 at max. So in a worst-case scenario it would become non-available within some months time from now, according to what was reported in 2008.
    On the other hand, A-Eon reported on December 21st 2010 that "Apple have now closed down the PA Semi product line", whatever that may mean specifically.

    > Nor have they considered licensing it out.

    That's not true. It was reported that Apple might try to sell the PA6T design after backing off from the chip. This would require a buyer, though.
  • »08.03.11 - 21:46
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4977 from 2009/1/28
    From: Delaware, USA
    Interesting, I've always looked at the PA6T as a more powerful processor intended for Apple laptops (as the G5 was just too power hungry).
    I haven't heard any reports that Apple was looking for a buyer for the PA6T, but as you pointed out they'd need an interested party.

    If Apple has ceased production then the product will only be available till supplies run out. This is not a good situation for A-eon.

    On paper, the design is impressive. I wonder what problems have kept it from being released to the beta testers?
    "Never attribute to malice what can more readily explained by incompetence"
  • »08.03.11 - 23:52
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12058 from 2003/5/22
    From: Germany
    > I've always looked at the PA6T as a more powerful
    > processor intended for Apple laptops

    Yes, that's how it started internally at P.A.Semi in 2003, before Apple announced the switch to Intel in June 2005, and even longer before it was introduced to the public in October 2005 as a chip for embedded use (due to lack of Apple target market).

    > I haven't heard any reports that Apple was looking
    > for a buyer for the PA6T

    With that I was still referring to July 2008's reports, which I'm sure you remember as we discussed them several times already:

    https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=7001&forum=3&post_id=72415#72415

    However, I don't know if that's still the plan.
  • »09.03.11 - 00:34
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4977 from 2009/1/28
    From: Delaware, USA
    >P.A. officials did not identify Apple but said the acquiring company may be willing to supply the chip on an end-of-life basis, if it could successfully transfer a third-party license to the technology

    OK, so Apple did appear willing to allow a third party to continue production. That qualifier about the transferability of the IBM license is daunting.
    Its a pity no existing company with an IBM license offered to continue manufacture. APM could have considered it.
    "Never attribute to malice what can more readily explained by incompetence"
  • »09.03.11 - 01:06
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4977 from 2009/1/28
    From: Delaware, USA
    I wonder how well the T5020 would work in a similar design. I still haven't gotten a complete idea of how many PCIe lanes the e5500 based processors support.
    "Never attribute to malice what can more readily explained by incompetence"
  • »09.03.11 - 01:10
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4977 from 2009/1/28
    From: Delaware, USA
    I'm not one for unsupported conspiracy theories, but isn't it odd that two of Apples recent acquisitions eliminated two promising PPCs. First Apple acquires PA Semi and phases out the PA6T, then they purchase Intrinsity and prevents APM from introducing Titan.
    I can't picture Steve Jobs making these acquisitions solely to damage the Power processor industry, but that is part of what has happened.
    "Never attribute to malice what can more readily explained by incompetence"
  • »09.03.11 - 04:46
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12058 from 2003/5/22
    From: Germany
    > OK, so Apple did appear willing to allow a third party to continue production.

    You're referring to the wrong part of the story. Your quote talks about the prospects of Apple supporting the chip and mentions the requirement for a transfer of P.A.Semi's Power Architecture license (obtained from IBM) from P.A.Semi to Apple, not from Apple to someone else. So, this has nothing to do with what would happen to the PA6T after Apple.
    The part I was referring to is this:

    "The letter suggests Apple will explore selling the designs to a third party after that time."

    As soon as Apple sells the chip, there wouldn't be any need for them to be "willing to supply the chip on an end-of-life basis" anymore.

    > That qualifier about the transferability of the IBM license is daunting.

    You're misreading that part. It refers to what was said 3 months before ("At that time...") by P.A.Semi. When the article was written in July 2008 the acquisition was already accomplished. And the fact that the PA6T started being produced again proves that P.A.Semi's Power Architecture license could successfully be transferred to Apple. I don't see a reason why this license couldn't be transferred from Apple to the PA6T's new owner (should there ever be any), then.

    > Its a pity no existing company with an IBM license offered to continue manufacture.

    As I said, history has proven that this license is transferable, so the new owner wouldn't be required to have one beforehand. Second, we don't really know if Apple is still having the chip manufactured or not. Third, we don't know if there's any negotiation between Apple and a potential new owner already going on.
  • »09.03.11 - 11:29
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12058 from 2003/5/22
    From: Germany
    > I wonder how well the T5020 would work in a similar design.

    You mean together with an AMD southbridge? Let's assume the T5020 is essentially an AltiVec-enabled P5020, having available the same, if not better/more, on-chip controllers as P5020. Would you really need a southbridge at all, then? The PA6T is a rather rudimentary SoC, whereas the P5020 is way more complete. So in summary, a T5020 based board design wouldn't be similar to a PA6T based board design.

    > I still haven't gotten a complete idea of how many PCIe lanes the
    > e5500 based processors support.

    For the P5020/P5010 Freescale says "4x PCI Express 2.0 controllers" (PCIe 2.0 is double the speed of PCIe 1.0, which is on the PA6T). Furthermore, they mention 18 SerDes lanes overall (PA6T has 24, but half-speed each), but I don't know if they're configurable like on the PA6T. On the other hand, more on-chip controllers means less PCIe lanes needed.
  • »09.03.11 - 12:04
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4977 from 2009/1/28
    From: Delaware, USA
    Quote:


    Andreas_Wolf wrote:
    > I wonder how well the T5020 would work in a similar design.

    You mean together with an AMD southbridge? Let's assume the T5020 is essentially an AltiVec-enabled P5020, having available the same, if not better/more, on-chip controllers as P5020. Would you really need a southbridge at all, then? The PA6T is a rather rudimentary SoC, whereas the P5020 is way more complete. So in summary, a T5020 based board design wouldn't be similar to a PA6T based board design.

    > I still haven't gotten a complete idea of how many PCIe lanes the
    > e5500 based processors support.

    For the P5020/P5010 Freescale says "4x PCI Express 2.0 controllers" (PCIe 2.0 is double the speed of PCIe 1.0, which is on the PA6T). Furthermore, they mention 18 SerDes lanes overall (PA6T has 24, but half-speed each), but I don't know if they're configurable like on the PA6T. On the other hand, more on-chip controllers means less PCIe lanes needed.


    4X isn't the most promising connection for a GPU although it would make a good connector for a Southbridge.But, as you've pointed out, many of the SC functions are incorporated into the P5020. I wonder if there is any way to utilize the SerDes lanes?

    Edit - Just reviewed the chip again. All SerDes lanes are assigned functions so the limit is 4x PCIe. The PA6T definitely has an advantage here.

    [ Edited by Jim on 2011/3/9 19:39 ]

    [ Edited by Jim on 2011/3/9 20:11 ]
    "Never attribute to malice what can more readily explained by incompetence"
  • »09.03.11 - 18:36
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12058 from 2003/5/22
    From: Germany
    > 4X isn't the most promising connection for a GPU

    I don't read this spec as the amount of lanes but as the amount of controllers. Look at the block diagram:

    http://www.freescale.com/files/graphic/block_diagram/P5020_BD.jpg

    There you have (from left to right) the following controllers:
    one 10GbE, five 1GbE, four PCIe, two SRIO, two SATA2, one Debug.

    > I wonder if there is any way to utilize the SerDes lanes?

    Below the controllers there's a block labelled "18-Lane 5 GHz SerDes". 5 GHz means 500 MB/s throughput per lane, btw. If you count the arrows from the controllers to the SerDes block you'll find that these don't amount to 18 but to only 12. To my mind, this seeming mismatch indicates that the controllers's assignment to the SerDes lanes may be configurable at the board designer's level, just like with the PA6T. So if the board designer thinks that the board could spare for instance the 10GbE port, three 1GbE ports, the two SRIO ports and the Debug port, he can distribute the 18 SerDes lanes across the remaining two 1GbE ports (would use 1 lane), the two SATA2 ports (would use 2 lanes) and would still have 15 lanes at disposal for the four PCIe ports, which could be configured as powering for instance one x8 slot (with x16 physical connector for graphics cards), one x4 slot, one x2 slot and one x1 slot. Let me reiterate that this is PCIe 2.0, so the x8 slot would provide the same bandwidth as the PCIe 1.0 x16 slot on the X1000.
    However, that's all assuming my idea that the SerDes lanes are configurable holds true.

    Edit:

    > All SerDes lanes are assigned functions

    You mean they're not configurable? That would be a pity. Do you know how many SerDes lanes are hardwired to which controller in detail? The block diagram isn't really helpful in this regard (see mismatch 12 arrows vs. 18 lanes).

    > so the limit is 4x PCIe.

    Which would still equal PCIe 1.0 x8 speed ;-)

    > The PA6T definitely has an advantage here.

    Definitely, provided your above assumption that the SerDes lanes are hardwired to the controllers is correct.

    [ Edited by Andreas_Wolf on 2011/3/9 21:42 ]
  • »09.03.11 - 19:31
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    amigadave
    Posts: 2793 from 2006/3/21
    From: Northern Calif...
    Quote:


    Jim wrote:
    I'm not one for unsupported conspiracy theories, but isn't it odd that two of Apples recent acquisitions eliminated two promising PPCs. First Apple acquires PA Semi and phases out the PA6T, then they purchase Intrinsity and prevents APM from introducing Titan.
    I can't picture Steve Jobs making these acquisitions solely to damage the Power processor industry, but that is part of what has happened.


    I don't subscribe to conspiracy theories either, but it is not outside the realm of possibilities that preventing any kind of resurgence, or popular success of any manufacturer trying to use any PPC architecture in a personal computer would be sabotaged by Apple to prevent any backlash or second guessing of their decision to move away from PPC CPUs in all their products. I doubt that any of Apple's decisions for these acquisitions was based only on any fear of a resurgence of PPC, but it might have been a tiny partial added benefit to their bigger plans to shut down certain PPC projects.

    Then again, it may have never crossed their minds and be just a coincidence. No way to ever confirm one way or the other.
    MorphOS - The best Next Gen Amiga choice.
  • »09.03.11 - 23:23
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4977 from 2009/1/28
    From: Delaware, USA
    >I don't read this spec as the amount of lanes but as the amount of controllers. Look at the block diagram:


    Then you read it the same way I do, which leaves us at the same point - not knowing how many PCIe lanes the processors support.
    "Never attribute to malice what can more readily explained by incompetence"
  • »09.03.11 - 23:51
    Profile