Crowdfunding for TALOS Workstation
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4857 from 2009/1/28
    From: Delaware, USA
    >Why would AROS need the Talos II for this?

    It doesn't.
    Currently I'm looking at ports for everything from the WiiU to PowerMac G4, G5s, iBooks and PowerBooks.
    The Power 9 platform is just the most capable PPC system, and the Lite version is actually affordable.

    SAM460 emulation versus X5000 emulation? The latter might be more capable, and as you pointed out, it could support little endian applications.

    The future looks promising.
    "Never attribute to malice what can more readily explained by incompetence"
  • »18.05.18 - 01:37
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 10585 from 2003/5/22
    From: Germany
    > the Lite version is actually affordable.

    ...but offers only one free PCIe slot after plugging a graphics card.

    >> In terms of MorphOS, will a virtualized X5000 have any advantage
    >> over a virtualized Sam460?

    > SAM460 emulation versus X5000 emulation? The latter might be more capable

    How so?

    > and as you pointed out, it could support little endian applications.

    ...which would require a little-endian PPC OS to run in between. MorphOS and OS4 are strictly big-endian, and a potential little-endian AROS/PPC wouldn't make sense running trapped inside the virtualization of another platform when it can run on POWER9 natively.
    Besides, I think that all Book III-E CPUs (which includes PPC4xx) can run in little-endian mode.
  • »18.05.18 - 13:03
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4857 from 2009/1/28
    From: Delaware, USA
    >> the Lite version is actually affordable.

    >...but offers only one free PCIe slot after plugging a graphics card.

    Which is an X4 slot that might potentially be able to be divided into four X1 slots.
    If that can be done, we have a slot for a network card, sound card, and USB 2.0 card (since the TalosII only supports USB 3.0 - but then again, as per another thread, I'd love to see us support USB 3.0 as it would greatly improve i/o throughput for external drives), leaving one X1 expansion slot for any other needed device.

    >> SAM460 emulation versus X5000 emulation? The latter might be more capable

    >How so?

    Simple, the SAM460 has a weaker cpu (even if we compare them at similar clock speeds) to the X5000.
    It also only devotes half as many PCI-e lanes to the video card as the X5000.
    If the emulation truly is accurate in its duplication of each boards capabilities, then obviously the X5000 emulation would be the more capable one.

    Of course, the best resolution to this would lay in just porting the OS to Power 9 natively.
    But even if this is never done, it still might be possible to run MorphOS with an X5000 or SAM460 emulation.
    On a system that is powerful enough to do a competent emulation of X64.

    Consider this for a second. A system that could run MorphOS PPC without resorting to qemu, and still run MorphOS X64 (or as I like to think of it MorphOS NG) via whatever emulation IBM is using for X64 (be it qemu or whatever).

    This is another reason for me to consider the TalosII Lite. I could still run Power based OS' and software, while running other OS sessions big or little endian, native or emulated, via the hypervisor.

    Further, what outside of legacy Amiga compatibility prevents us from creating a little endian variant of MorphOS?

    And wouldn't a port to X64 be easier if you could run BOTH forks of MorphOS on the SAME machine?

    Think about it, the cheapest Power 9 cpu, the four core supports sixteen concurrent threads. That is as many as the BEST Ryzen 7 cpus (the 1700, 1800, 2700X, and the still unreleased 2800X).

    And Power 9 has variants that go up to 22 cores (at AMD, only Epyc can beat that core count, and a 32 core Epyc processor supports only 64 concurrent threads versus a 22 core Power 9 cpu which can handle 88 concurrent threads - double that for a two processor system and the Power 9 system would support a whopping 48 more threads that the Epyc system).

    >...and a potential little-endian AROS/PPC wouldn't make sense running trapped inside the virtualization of another platform when it can run on POWER9 natively

    Right, it WOULDN'T make sense because AROS has a switch that allows it to be compile as either big endian or little endian.
    So initially AROS will be run hosted on a big endian variant of Linux, the next goal being a move to a little endian variant of Linux, with a final goal of a native little endian port.

    This X64 fork that has already been committed IS a risky venture that might not bring us a significant increase in our user base, but until a qemu emulation for MorpOS PPC is adopted it WILL eliminate compatibility with our current base of software unless that software is recompiled. And legacy Amiga software will require either UAE or the afore to mentioned qemu emulation of MorphOS PPC.

    So, to state this clearly, I AM adopting Power 9. It should be capable of running BOTH forks of MorphOS. Obviously a powerful X64 system could do the same.
    But at around $2000, the Power 9 system is not significantly more expensive than a high end Ryzen based system, and it actually cost LESS than an X1000 did.

    I the past, we argued over adopting the X5000, decide not to adopt the 11,2 PowerMac even though in many ways it is superior to the X5000, and now we appear to be dismissing a system that runs faster both, is capable of running 8 times as many concurrent threads as the X5000 and 4 times as many as the Quad G5, with enough power to realistically consider emulating an X64 system (and the X5000 will never be able to do that adequately).

    Outside of WarpOS, MorphOS was the earliest operating system on the Amiga market to support PPCs (it might even predate WarpOS, I'm not sure).
    And for all the calls to drop PPCs in favor of X64, it means jettisoning our original ISA in favor of the ISA of the direct competitor to both the 68K and the PPC.
    And we should do this just because Apple did?
    Fuck Apple, they should have gone completely to ARM, and btw I would like to remind you that that was the ISA I was advocating before the decision was made to move to X64.

    So, I will follow you all to X64, I just might not be running the software on an X64, unless we face serious platform limitations.
    I which case I'll build an X64 system just for MorphOS.

    But I'm feed up with Microsoft, I never cared for Intel, and the idea of having an open system (with everything from the hardware, to the firmware, and the OS being open) is VERY attractive to me.

    So this Summer I'm experimenting with Linux hosted AROS ports to a variety of PPC platforms, with the eventual goal of porting it to Power 9 (and no doubt the T2080 laptop if it ever reaches production).
    I'll also stay informed about SAM460 and X5000 emulation, as well as X64 emulation on Power 9 platforms.

    My goals are set. The MorphOS developers goal to fork to X64 is set.
    But that doesn't mean that MorphOS PPC development has to stop.

    The AROS community doesn't feel obligated to limit themselves to one ISA.
    Why should we? Because of limited developer resources?
    Bullshit. Open up development to more programmers.

    So what if MorphOS is proprietary.
    Even Hyperion uses outside developers.
    That's what non-disclosure agreements are for.

    In short, let's stop dicking around, and instead of thinking of this as a limited hobbyist OS, how about considering the original goal which was to create a REAL alternative OS?

    I'm tired of the "Pinocchio" crap.

    And I believe you all think to small. I aim to punch up, not limit myself to slow incremental change.
    So I'm still a part of your community, but I had to join a Linux group to get the Qorlq platform I wanted created (the majority of the Amiga community being too timid to think it could be done, now that I've adopted Linux, I might as well explore AROS.
    And I want an alternative platform, not just another alternative OS in the great sea of options in the X64 world.

    I think the fixation many of you have on limiting our future direction(s) may eventually lead to us drown in that "sea".

    [ Edited by Jim 18.05.2018 - 11:11 ]
    "Never attribute to malice what can more readily explained by incompetence"
  • »18.05.18 - 15:56
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 10585 from 2003/5/22
    From: Germany
    >>> the Lite version is actually affordable.

    >>...but offers only one free PCIe slot after plugging a graphics card.

    > Which is an X4 slot

    It's an x8 slot.

    > that might potentially be able to be divided

    You mean with a PCIe splitter card?

    > the TalosII only supports USB 3.0

    It has one USB 2.0 port.

    > the SAM460 has a weaker cpu [...] to the X5000. It also only devotes
    > half as many PCI-e lanes to the video card as the X5000. If the emulation
    > truly is accurate in its duplication of each boards capabilities, then
    > obviously the X5000 emulation would be the more capable one.

    Both Cyrus/X5000 and Sam460 devote 4 PCIe lanes to the PEG slot. It's just that the lanes of Cyrus/X5000 are twice as fast (PCIe v2 vs. v1).
    But I don't think any of this makes a difference when emulating/virtualizing the systems with QEMU/KVM.

    > what outside of legacy Amiga compatibility prevents us from
    > creating a little endian variant of MorphOS?

    Source code access, and time and knowledge to fix all the big-endian assumptions? :-)

    >>> X5000 emulation [...] could support little endian applications.

    >> a potential little-endian AROS/PPC wouldn't make sense running
    >> trapped inside the virtualization of another platform when it
    >> can run on POWER9 natively

    > Right, it WOULDN'T make sense

    So which OS do you want to run on that virtualized X5000 running in little-endian mode?

    > Outside of WarpOS, MorphOS was the earliest operating system
    > on the Amiga market to support PPCs

    If you consider WarpOS, you should consider PowerUP as well, as it was the true predecessor to MorphOS, written by the same man.

    > it might even predate WarpOS, I'm not sure

    MorphOS (a complete PPC OS) is what emerged from the shortcomings of WarpOS and PowerUP (PPC kernels running the PPC as a mere co-processor to the m68k).

    > Open up development to more programmers.

    I think anyone can apply for MorphOS team membership and subsequent source code access. And if the powers that be feel that the applicant brings something tangible to the table I'm sure he or she won't be rejected.

    > Even Hyperion uses outside developers.

    Even? They simply cannot use any "inside delelopers" :-)
  • »18.05.18 - 17:55
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4857 from 2009/1/28
    From: Delaware, USA
    Well, this has been useful.
    I missed the USB 2.0 port, but we still would probably need a NEC based USB 2.0 solution.

    "Splitting" a PCI-e card slot isn't as simple as it sounds, but its one avenue to pursue.

    Since the Talos II can run big endian Linux distros, my first target would be hosted AROS, as the code is obtainable (actually thanks to a pointer from Nik, I already have it).

    Since SAM460 and X5000 emulation is under development, we'll have to see how that work on the PowerMac G5, but anything the PowerMac G5 can run ought to be able to be run on the TalosII.

    And I don't have the talent, or the Amiga background to consider applying for developer access.

    In fact, the last significant thing I did on a 68K platform was create a driver for Tseng Labs ET4000 based ISA video cards (and even that featured some cheats like hard coding the register addresses instead of accessing the X86 bios - this meant only about half the cards on the market would work with that driver without the creation of a modified version).

    But it did work, and to this day I'm rather proud out it, as we provide it to other users of Peripheral Technology's PT68K2, K4, and K5 boards (that last board isn't well known, probably because by that time we had been working directly with the hardware designer and this '020 variant was created at our request, along with two 6809 coprocessor cards, the second of which would have featured an mmu and vdg specifically for the 6809, and a V20 card).
    By the time we'd gotten to the 4, we were supplying complete end users systems, while Peripheral Technology focused on bare and populated board.

    Which made sense as we had the licensing agreement with Microware for reduced cost purchases of their OS and BASIC, and we owned the rights to the GUI we'd ported from the work of a programmer who'd originally written it for Gespac.

    Anyway, I digress. I can write C code, at one time I was fairly proficient with 6809 and 68000 assembly, but I don't have a good grasp of Python or MUI.
    And ALL the code I wrote was designed to be reentrant.

    I know that can be done under AROS and MorphOS, but I don't believe its mandatory.

    The only really talented programmers outside of the MorphOS are first, Staf Verhagen from the AROS community (who has dropped out of active development to pursue an interest in ASICs) and Hans de Ruiter, who as an OS4 developer might met some resistance even if he had the time to commit to additional projects. BUT, between Mark Olsen and Hans de Ruiter, I have learned a fair amount about modern gpu architecture.
    Considering my work with video display generators was limited to unaccelerated cards where the most important factor was how large the transfer rate from the cpu or memory, through the ISA bus, to the video card was.
    BTW - That's what the chips were consider VDGs, then as basic accelerators where introduced VDPs (video display processors), finally evolving into the term gpus (graphics processing units)

    Anyone that remembers that period will then understand why I insisted on using ET4000 based cards for our systems, in that these cards were faster than ANY other cards on the ISA market until the advent of acceleration (and frankly, were better than some of the cruder early accelerated cards).

    In any case, I'm not competent enough to do much more than compile existing Linux code and make slight alterations and modifications to an OS or maintain a system.

    So AROS is about my speed, especially if its hosted.

    And if an emulation of a supported MorphOS platform is complete, I'm reasonable sure I can get that running on a Power 9 platform.

    But contributing to or even having the hubris to suggest what the existing developers might do?
    Its way above my pay grade. These guys cut there teeth on Amigas and the Amiga OS API, the 68K API's I worked under were more Linux like (the primary OS being the as that used, with modifications, as an embedded OS for DVI disc players).
    Point of fact, except for the hard coded addressing of the ET4000 driver, nothing I ever did referenced specific address locations as the point of position independent code is that everything is relative.

    Frankly, I used to look down on some Amiga coders as they used tricks I would never have considered, like self modifying code, and for that matter any other stunt that might get them the result they were aiming for. Almost ALL of my interaction with hardware went through driver, nothing banged the hardware.
    After all, what if there were variations in the platform, or even in similar component.

    I know MorphOS coding does rely on these "dirty" trick, and that MorphOS can support reentrant and position independent code, but from what I've seen, it doesn't appear to require it.

    I moved from an environment certified for mission critical applications, to a general interest in 68K platforms, and then, thanks solely to MorphOS to an interest in PPCs.
    Oh, and btw, Apple's use of the 68K was incredibly primitive, so I'd disregarded them from the start, and never would have considered their PPC based systems with MorphOS (and its micro kernel) to motivate my interest.

    But guys, this is IT.
    We are now committed to an X64 future, and frankly as fond as I am of AMD, I frickin' hate Intel (and Microsoft? My ex-wife used to contend that Bill Gates was the anti-christ).

    Just this year I've experience stability issues with Win10, Win7, and Office (in particularly Excel, where I had two very weird issues that required a reboot, although that did occur in Word as well).
    I didn't think I could replace Excel (Word, hey I can use Open Office for that capability), but since my primary work with Excel is statistical, I've shifted to Minitab which while not as presentation oriented as Excel, is a MUCH better product for statistical analysis.
    As soon as I can coerce IBM into giving me a good price for a license, I'll be using SPSS as well.
    This latter package ought to be available under Linux (considering IBM's support for the OS), I'm unsure about Minitab (I'll check right after this post).

    SO, as I've said before, I'm incredibly tired of the Wintel platform, I don't want to move to MacOS and get raped on hardware pricing, so I AM going to check this out.
    I've lived with Linux before, I think I can make do with it, and if I need Office (Minitab spreadsheets can be transferred to and from Excel if there is an Excel feature I HAVE to have), I can't running it an emulated X64 session on a Power 9 platform.

    I can't dictate where you all are going, AND for that matter I think X64 was a pretty good idea.
    I've mentioned this to Andreas, and I don't think Bill Buck would be offended if I posted this part of his last e-mail, but Bill's had recent conversations with the director in charge of the division at NXP that oversee PPCs (as well as a lot of other devices) and the PPC IS dead at NXP.
    Simply stated, there is no intention to develop that ISA any further, products that are slated for long term support will remain available.
    But this is a dead end at NXP, it likely is at AMCC as well, and the only real flicker of life is in the embedded products IBM designs and Power 8, 9, and beyond.

    So the latter day inheritor of Phillip's legacy (Phillips once being a licensee and second source to Motorola) has effectively pulled the plug on the last ISA that Motorola was involved in developing and Freescale continued forward with.
    Face it, we knew this was inevitable when they canceled the e700 core, and while the e5500 and e6500 cored products do have some advanced features, they weren't really aimed at the desktop market (and btw, the e500v2 is REALLY unsuited for that market).

    I completely support the developers decision to fork to X64 (especially if we can support AMD components), but at the same time, that is going to make us "yet another X86/X64 operating system", one of thousands dominated by two companies Intel and Apple that I frankly despise.

    If the hardware support is so restrictive under MorphOS X64 that I have to buy specific X64 hardware in order to run it, I will as long as I'm not forced to buy Intel based hardware.
    I have enough of that already. I am willing to commit an AMD based system to the OS, but I'm not loading Windows on it.

    And I'd like to see MorphOS PPC continue, if only on the back burner. With that in mind, I'll be retaining a handful of registered Apple hardware.
    And were that to occur, some backporting of MorphOS NG/X64 makes sense.
    The PowerMac G5 can handle a 64 bit OS, and I've seen modification that allow memory addressing up to 64 GB (depending on the model).
    And Power 9 system can do EVERYTHING that our new ISA can do, and in a few areas its BETTER.
    You can't build a desktop Ryzen system with 22 cores unless you rely on the enterprise level multi-die variant.

    A Talos II can be be started with one 4 core cpu that supports 16 concurrent threads (as many as the most capable single die Ryzen cpus), but that same board will accept all the other Power 9 variants right up to the 22 core version which would support a frightening 88 simultaneous threads.

    And if you stick with the standard Talos II instead of the Talos II Lite, a dual cpu system could built that started out with that same single 16 thread cpu, but could be expanded to a dual cpu platform supporting as many as 176 simultaneous threads.

    Know any end user boards in the X64 market that can do that?

    So is Power dead? At NXP it is, and I'm still working with a team aiming to complete a T2080 based laptop design, I'd also still advocate that Aeon rethink its low end strategy and build a starter board with a T10XX cpu.
    Also, I now have the full schematics and board layouts for the RDB4240, and that has a ridiculous core/thread count.

    BUT, Power...its all Power 9 now really. And its affordable.
    As mentioned, I can't dictate where you all go, but I'm supporting it.
    And I believe it would make an excellent transition platform for the next step in MorphOS evolution, since it could run PPC variants of our OS' natively and X64 variants under emulation (just the opposite of what our future X64 platforms will have to do, MorphOS NG/X64 natively, and MorphOS PPC via qemu).

    Either way, I'm not pissing on my legacy.
    Intel products were dreadful during the period I worked in 68K development, just because they had an idiotic number of revisions to improve, doesn't mean that there is nothing to challenge them.

    There's ARM (which we've summarily dismissed), and there's still Power (which I think will be capable enough to run any X64 code developed).

    So again, am staying right HERE. I'm going to work to broaden the PPC Platforms supported under AROS, then I'm going to look at a port to Power 9.
    And if there's no support for 11,2 PowerMacs or Talos II based systems, I'm still going to keep using them.

    I'm kind of hard headed like that once I've reached a conclusion.

    [ Edited by Jim 18.05.2018 - 17:52 ]
    "Never attribute to malice what can more readily explained by incompetence"
  • »18.05.18 - 22:29
    Profile
  • Moderator
    Kronos
    Posts: 1908 from 2003/2/24
    Bout the USB, USB 1.1 worked on 2.0 ports long before we had support for 2.0.

    AFAIK thats also true for 2.0 on a 3.x port.

    We are also not limited to NEC chipsets its just that these seem to be more reliable than VIA.
    --------------------- May the 4th be with you ------------------
    Mother Russia dance of the Zar, don't you know how lucky you are
  • »18.05.18 - 23:20
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4857 from 2009/1/28
    From: Delaware, USA
    Quote:

    Kronos wrote:
    Bout the USB, USB 1.1 worked on 2.0 ports long before we had support for 2.0.

    AFAIK thats also true for 2.0 on a 3.x port.

    We are also not limited to NEC chipsets its just that these seem to be more reliable than VIA.


    Agreed, I just installed a Renesas UDP720202 based USB 3.0 Express card in my i5 laptop, and once I found the proper drivers (on Renesas' website, doh) its been running completely perfectly.

    And Renesas owns NEC and its IP, so any current production of the NEC USB 2.0 chips we rely on comes from Renesas, not NEC.
    I've been keeping track of Renesas, because I believe they also adsorbed Hitachi.
    Its still a relatively unknown name, but they have a nice solid collection of IP.

    So you think we might initially be able to at least get USB 2.0 performance out of USB 3.0 ports?
    That could prove useful on systems that have limited USB 2.0 ports.

    For example, I was under the impression that Raptor Engineering TalosII board only had USB 3.0 ports, until Andreas informed me today that it has one USB 2.0 port.
    You could use a hub to expand that, but sometimes that causes issues.

    Being able to use USB 3.0 hardware at USB 2.0 speeds would be a good starting point for the eventual support of full USB 3.0 speeds.

    And believe me, that 3 to 4 fold improvement in throughput really makes a difference.

    I was using a USB 3.0 SATA adapter with a 2.5" SATA150 laptop drive on the afore to mentioned i5 based laptop connected to a USB 2.0 port.
    The improvement in performance with the USB 3.0 card is just astounding.

    I've never used this interface for backups before because it was just too damned slow, with this upgrade it becomes a practical option.

    [ Edited by Jim 18.05.2018 - 18:50 ]
    "Never attribute to malice what can more readily explained by incompetence"
  • »18.05.18 - 23:47
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 10585 from 2003/5/22
    From: Germany
    > X5000 emulation is under development

    Is it? It's known who works on Sam460 emulation, but who works on X5000 emulation?

    > MorphOS coding does rely on these "dirty" trick

    Huh?

    > if I need Office [...], I can't running it an emulated X64 session on a Power 9 platform.

    ;-)

    > PPC [...] is a dead end [...] likely [...] at AMCC as well

    Applied Micro ceased existence a while ago. And yes, PPC is a dead end at MACOM, the new owner.

    > the only real flicker of life is in the embedded products IBM designs and
    > Power 8, 9, and beyond.

    Embedded Power Architecture from IBM? Has there been anything after the A2-based chips from 2010/2011 and the 476FPE and 476GTR from 2011/2013 (and Espresso from 2012, if we want to include this)?

    > that is going to make us "yet another X86/X64 operating system", one of thousands
    > dominated by two companies Intel and Apple

    Thousands? I don't think Apple dominates any OS but its own.

    > The PowerMac G5 can handle a 64 bit OS, and I've seen modification that
    > allow memory addressing up to 64 GB (depending on the model).

    Any details?
  • »19.05.18 - 13:52
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 10585 from 2003/5/22
    From: Germany
    > Renesas owns NEC and its IP

    Just NEC Electronics, which was a spin-off of the semiconductor operations of NEC.

    > Renesas [...] I believe [...] also adsorbed Hitachi.

    It's more like the other way round. Renesas started as a joint venture between Hitachi and Mitsubishi.
  • »19.05.18 - 14:09
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4857 from 2009/1/28
    From: Delaware, USA
    Andreas,

    the statement about MorphOS should has contained a "not", sorry.

    And as I've been using Hitachi components since the HD6309 was introduced, I'm fully aware of how Renesas evolved.

    In any case, was I wrong in stating that they own Hatachi and NEC's electronics IP?
    Or did you misinterpret my statement about NEC to mean any other part other than their electronics division (the only division I have any interest in).

    In any case, could we get back to the subject at hand?

    Raptor's specifications state that there are two USB 2.0 connectors on the Talos II Lite case.

    I haven't found a device that can properly break an X8 PCI-E slot into X4 slots (or what I would prefer, 1 X4 and 4 X1 slots).
    But I have found a company that has an X4 host interface that will properly connect to four seperate X1 slots.
    And there are other devices, but I need to figure out if PLX switching is required, or if the lanes can simply be separated via routing.

    In any case, since the Talos II supports sound output via hdmi on Radeon HD video cards, a sound card isn't actually required.
    There are two front panel USB 2.0 connectors on the Talos II Lite chassis, and those could be expanded with hubs, or as Kronos has suggest the USB 3.0 connections could be used.
    Networking? Well, there are two Broadcom Gigabit ethernet ports on board, so instead on resorting to a card with an currently supported network interface, it might just make more sense to put in the effort to create a native NIC driver.

    So, I'm not sure I'd even need to use the X8 PCI-E slot, and if additional expansion is required, it ought to be possible to and hardware to the X8 slot that expands it to provide the connectivity for X1 sound, network, and USB 2.0 cards.
    "Never attribute to malice what can more readily explained by incompetence"
  • »19.05.18 - 18:40
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 10585 from 2003/5/22
    From: Germany
    > was I wrong in stating that they own Hatachi and NEC's electronics IP?
    > Or did you misinterpret my statement about NEC to mean any other part
    > other than their electronics division

    As I read those statements ("owns NEC", "adsorbed Hitachi"), they could be understood as meaning that Renesas is the parent company or owner of NEC and/or Hitachi, when what Renesas owns is just the IP of the (former) semiconductor operations of both.

    > Raptor's specifications state that there are two USB 2.0 connectors on the
    > Talos II Lite case. [...] There are two front panel USB 2.0 connectors on the
    > Talos II Lite chassis

    Yes, odd, two ports on the case but only one port according to the specs further down the page.

    > since the Talos II supports sound output via hdmi on Radeon HD video cards,
    > a sound card isn't actually required.

    This feature probably needs OS support more than it needs support by the Talos II ;-)
  • »20.05.18 - 19:17
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4857 from 2009/1/28
    From: Delaware, USA
    >This feature probably needs OS support more than it needs support by the Talos II ;-)

    Yeah, I want to talk to Mark about that, but I've bugged him enough for one weekend.

    On the dual USB 2.0 ports, the case probably just features a hub, thanks for pointing out that the board had USB 2.0.
    "Never attribute to malice what can more readily explained by incompetence"
  • »20.05.18 - 19:36
    Profile
  • Moderator
    Kronos
    Posts: 1908 from 2003/2/24
    Not sure what you are talking bout Renesas,NEC or Hitachi and their driver support.

    The issue with USB2 is that the chips still had all the registers used for USB1.1 just like a VGA card could still do CGA or MDA.

    No idea wether thats still true for USB3 vs USB2.
    --------------------- May the 4th be with you ------------------
    Mother Russia dance of the Zar, don't you know how lucky you are
  • »20.05.18 - 19:50
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4857 from 2009/1/28
    From: Delaware, USA
    Quote:

    Kronos wrote:
    Not sure what you are talking bout Renesas,NEC or Hitachi and their driver support.

    The issue with USB2 is that the chips still had all the registers used for USB1.1 just like a VGA card could still do CGA or MDA.

    No idea wether thats still true for USB3 vs USB2.


    Good question, but isn't it safe to assume that USB 3.0 has register level compatibility with USB 2.0?
    "Never attribute to malice what can more readily explained by incompetence"
  • »20.05.18 - 20:55
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4857 from 2009/1/28
    From: Delaware, USA
    You da man, Andreas.

    I need this kind of stuff.

    BTW, anything similar for the NXPT4240?
    In particular, can the SerDes lanes assigned to network connectivity be assigned to other devices?
    "Never attribute to malice what can more readily explained by incompetence"
  • »21.05.18 - 16:18
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 10585 from 2003/5/22
    From: Germany
    > anything similar for the NXPT4240? [...] can the SerDes lanes assigned to
    > network connectivity be assigned to other devices?

    Except by selecting another predefined SerDes config that assigns less SerDes lanes to the network controllers and more to other devices, no.
  • »21.05.18 - 18:30
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 10585 from 2003/5/22
    From: Germany
    > the summary of that seems to state that the X8 slot could be bifurcated
    > but the primary reason its problematic is that those lanes also control
    > another device. I could use some information on that device and which
    > lanes it is connect to (some or all).

    As I understand it, the x8 slot on the Talos II Lite comes from an x16 controller that already bifurcates into this x8 slot and a second x8 connection to the SAS controller. The x8+x8 bifurcation of this x16 controller is fixed in silicon and thus cannot be changed.
  • »21.05.18 - 19:15
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4857 from 2009/1/28
    From: Delaware, USA
    Quote:

    Andreas_Wolf wrote:
    > the summary of that seems to state that the X8 slot could be bifurcated
    > but the primary reason its problematic is that those lanes also control
    > another device. I could use some information on that device and which
    > lanes it is connect to (some or all).

    As I understand it, the x8 slot on the Talos II Lite comes from an x16 controller that already bifurcates into this x8 slot and a second x8 connection to the SAS controller. The x8+x8 bifurcation of this x16 controller is fixed in silicon and thus cannot be changed.


    Not even using a PLX switch?
    And the SAS controller is optional, the board appears to be available without it.
    "Never attribute to malice what can more readily explained by incompetence"
  • »21.05.18 - 20:20
    Profile
  • Caterpillar
    Caterpillar
    xilinder
    Posts: 39 from 2018/2/1
    From: USA
    >The x8+x8 bifurcation of this x16 controller is fixed in silicon and thus cannot be changed.<

    That is also how I understood the PCIe breakout from the CPU, hence the dual cpu board is what I ordered.

    I know it's more expensive, but for the future, perhaps a single cpu on a dual board is a better buy.
  • »21.05.18 - 21:23
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 10585 from 2003/5/22
    From: Germany
    >> The x8+x8 bifurcation of this x16 controller is fixed in silicon
    >> and thus cannot be changed.

    > Not even using a PLX switch?

    I think using a PCIe switch on the x8 slot could work as this has nothing to do with the bifurcation of the POWER9's PCIe controller.

    > the SAS controller is optional, the board appears to be available without it.

    I guess the lanes dedicated to this SAS controller are routed to the footprint no matter if populated or not.
  • »21.05.18 - 23:13
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4857 from 2009/1/28
    From: Delaware, USA
    Quote:

    Andreas_Wolf wrote:
    > anything similar for the NXPT4240? [...] can the SerDes lanes assigned to
    > network connectivity be assigned to other devices?

    Except by selecting another predefined SerDes config that assigns less SerDes lanes to the network controllers and more to other devices, no.


    Yeah, I'm trying to figure out the available configurations.
    There appears to be a way to configure more PCI-E lane than the X8, X4 combo used on the T4240RDB-PB (at least 16 lanes, possibly 17), but I'm not sure that that doesn't disable SATA.
    One I2C lane, I2C4, isn't even used on that reference board, and other SerDes lane groups appear not to be fully utilized.
    It going to take a little more examination of the SOC vs the board documentation to sort this out.
    "Never attribute to malice what can more readily explained by incompetence"
  • »22.05.18 - 12:09
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4857 from 2009/1/28
    From: Delaware, USA
    Andreas, you are the man, your recall saves me so much time.

    Thanks for those two references.

    So,if available configurations include

    1. x8-gen2 x8-gen2
    2. x8-gen2 x4-gen3 x4-gen2
    3. x4-gen3 x4-gen2 x4-gen2 x4-gen2

    Wouldn't option 3 make sense since an X4 gen 3 slot has the same bandwith as an X8 gen 2 slot?

    And that would mean that the T4240QDS likely uses this last option?
    If the T4240RDB-PB uses something similar to option 2, is it utilizing gen3 lanes on the single X4 slot, and what about those two remaining PCI-e lanes even if SATA is retained?

    What does that leave?

    X8 gen2, X4 gen3 (or2) with two lanes unused.
    Or, X4 gen3, X4 gen2, X4 gen2 again with two lanes unused (and two reserved for SATA 2).

    Seems like not utilizing the onboard SATA controller and going with an X4 gen3 (to supply an X16 slot) for the video card, with two X4 gen2 slots and four lanes serving an SATAIII controller would make more sense.

    Three slots, instead of two, and the same bandwidth for video.

    [ Edited by Jim 22.05.2018 - 17:06 ]
    "Never attribute to malice what can more readily explained by incompetence"
  • »22.05.18 - 21:32
    Profile