AROS for 68k question
  • Acolyte of the Butterfly
    Acolyte of the Butterfly
    Posts: 129 from 2015/8/23
    As You may easily find Vampire with core5 in sysspeed is still slower than 68060.
    Especially in higher resolutions.
    Vampire with core5 get 95,73 MIPS while 68060 80 MHz get 106,10 MIPS.
    68k "altivec" is interesting but for now there is no software which use it.
  • »09.06.16 - 07:36
    Profile
  • Acolyte of the Butterfly
    Acolyte of the Butterfly
    Posts: 129 from 2015/8/23
    Quote:

    Yasu wrote:
    The point is that since it will be incompatible anyway, why keep all the legacy problems too?



    This is very good summary of the problems with aros x86.
    Aros x86 is not source and binary compatible and has all problems which amiga os has like lack of memory protection, resource tracking, suport for multicore.
  • »09.06.16 - 07:40
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 9285 from 2003/5/22
    From: Germany
    >> The point is that since it will be incompatible anyway,
    >> why keep all the legacy problems too?

    > This is very good summary of the problems with aros x86.
    > Aros x86 is not source [...] compatible

    As I already told you in comment #42 a week ago, AROS (no matter the ISA) is source-compatible with AmigaOS 3.1.

    > and has all problems which amiga os has like lack of memory
    > protection, resource tracking, suport for multicore.

    As I already told you in comment #42 a week ago, that's because it is source-compatible with AmigaOS 3.1.
  • »09.06.16 - 08:54
    Profile
  • Acolyte of the Butterfly
    Acolyte of the Butterfly
    Posts: 129 from 2015/8/23
    AROS is not source-compatible with AmigaOS 3.1.
    You have to spend some time coding for aros x86 and You get that most functions from amiga os are not compatible with amiga os because they were never tested with 68k software.
    Especially ZUNE is worth nothing crap which is far away from MUI 3.8 from 1997.
  • »09.06.16 - 09:07
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 9285 from 2003/5/22
    From: Germany
    > AROS is not source-compatible with AmigaOS 3.1.

    It may not be 100% compatible at this point. Actually, it's 82% compatible with AmigaOS 3.1 according to http://aros.sourceforge.net/introduction/status/everything.php (as of 6 years ago), but it clearly aims at full source compatibility, hence the restrictions.

    > You get that most functions from amiga os are not compatible with amiga os

    I've a hard time believing this "most functions" (i.e. >50%) claim (assuming that one of the two "amiga os" occurences means "aros).

    > because they were never tested with 68k software.

    Isn't this what OlafSch and wawa have been doing for years?

    > ZUNE is worth nothing crap which is far away from MUI 3.8 from 1997.

    MUI has not been part of AmigaOS 3.1, so Zune's MUI compatibility status has no bearing on AROS' source compatibility with AmigaOS 3.1.
  • »09.06.16 - 09:29
    Profile
  • Order of the Butterfly
    Order of the Butterfly
    OlafSch
    Posts: 161 from 2011/11/16
    Quote:

    ppcamiga1 schrieb:
    AROS is not source-compatible with AmigaOS 3.1.
    You have to spend some time coding for aros x86 and You get that most functions from amiga os are not compatible with amiga os because they were never tested with 68k software.
    Especially ZUNE is worth nothing crap which is far away from MUI 3.8 from 1997.



    Aros is source compatible, there are small differences on library level like missing or unimplemented fuctions but also new ones compared to 3.5 f.e. but it is very compatible. The 68k branch also is binary compatible and MUI 3.8 and even the newer one (however you call it) can be easily used by copying. Developers are indeed testing it together with Aros 68k and compatibility has improved considerably in recent time. Your posting does not sound like you having any experience with it so where do you know current state of Zune?

    [ Editiert durch OlafSch 09.06.2016 - 11:59 ]
  • »09.06.16 - 09:30
    Profile
  • Order of the Butterfly
    Order of the Butterfly
    OlafSch
    Posts: 161 from 2011/11/16
    Quote:

    ppcamiga1 schrieb:
    As You may easily find Vampire with core5 in sysspeed is still slower than 68060.
    Especially in higher resolutions.
    Vampire with core5 get 95,73 MIPS while 68060 80 MHz get 106,10 MIPS.
    68k "altivec" is interesting but for now there is no software which use it.





    nobody (or at least almost nobody :) ) realistic expects such a solution outperforming modern hardware. I do not know where you get your numbers, the last information I had wad that it runs comparable to 68060 100 Mhz. Also it includes lots of new commands so software can be optimized and run faster on it. Yes it is a toy and I do not expect it to overrun the market, it is filling a need in the 68k community where people wanted better hardware for their aging amigas and that at least affordable. So it fills the gap there but 68k is still way faster running in UAE on modern hardware. Amiga today is a retro market mainly with at least parts of the community being interested in better software and that needs truecolor, more processing power and more ram (partly much more ram). BTW at the moment all PPC options are toys compared to modern platforms because of no up-to-date browser and no Javascript JIT even beaten by simple smartphones.

    [ Editiert durch OlafSch 09.06.2016 - 12:10 ]
  • »09.06.16 - 09:41
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 9285 from 2003/5/22
    From: Germany
    >>> vampire [...] is still slower than 68060 80 MHz

    >> That's not true from what I can see from reports and videos. Vampire V2
    >> seems about as fast as an accelerator with 120 to 175 MHz 68060.

    > Vampire with core5 in sysspeed [...] get 95,73 MIPS while
    > 68060 80 MHz get 106,10 MIPS.

    SysSpeed's MIPS "benchmark" seems flawed, as are most MIPS ratings actually. What are the results of the other SysSpeed benchmarks?
  • »09.06.16 - 09:42
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 9285 from 2003/5/22
    From: Germany
    > I do not know where you get your numbers

    He wrote where he got (i.e. selectively chose) them from :-)
  • »09.06.16 - 09:54
    Profile
  • Order of the Butterfly
    Order of the Butterfly
    OlafSch
    Posts: 161 from 2011/11/16
    I do not know how realistic all these benchmarks are. But if it now performs like 68060 80 Mhz or 100 Mhz most amiga users (here I mean the old amigas) do not own one of the rare expensive old accellerators so if you want a upgrade card that is not costing a fortune (if you find one on ebay or other sites at all) you do not have any other realistic chance. It is already offering fast processor, 128 MB and RTG, I think planned is also chipset (SAGA) and better sound.

    [ Editiert durch OlafSch 09.06.2016 - 12:08 ]
  • »09.06.16 - 10:07
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4221 from 2009/1/28
    From: Delaware, USA
    The MIPS figure for the 68060 are still pretty impressive.
    I have a project I am working on (not Amiga related) that uses a 20 MHz 68HC000 and the MIPs/MHz ratio is a fraction of the '060's figure.

    But since I need the E output clock, an overclocked 68HC000 is the best option for me.
    "Magnetic was troubled by my avatars and 'satanic' references" - Jim Igou

    "Never attribute to malice what can more readily explained by incompetence"
  • »09.06.16 - 13:33
    Profile
  • Acolyte of the Butterfly
    Acolyte of the Butterfly
    Posts: 129 from 2015/8/23
    Sysspeed is well known, reliable benchmark.
    Sysspeed works on almost every amiga configuration.
    There ara many sysspeed results called modules on aminet.
  • »09.06.16 - 16:51
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 9285 from 2003/5/22
    From: Germany
    >> What are the results of the other SysSpeed benchmarks?

    > There ara many sysspeed results called modules on aminet.

    Unless there're also ones with Vampire2/Apollo results, your answer to my question doesn't make sense.
  • »09.06.16 - 17:38
    Profile
  • Acolyte of the Butterfly
    Acolyte of the Butterfly
    Posts: 129 from 2015/8/23
    "Happy" users of vampire post many screenshots from syspeed working on vampire.
    Poor performance of vampire is no secret.
    You just have to find it.
  • »09.06.16 - 17:53
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 9285 from 2003/5/22
    From: Germany
    > "Happy" users of vampire post many screenshots from syspeed working on vampire.
    > Poor performance of vampire is no secret. You just have to find it.

    Why so secretive? Just give links to some of these "many screenshots" showing also SysSpeed benchmarks other than MIPS for Vampire2/Apollo.
  • »09.06.16 - 21:53
    Profile
  • Acolyte of the Butterfly
    Acolyte of the Butterfly
    Posts: 129 from 2015/8/23
    It is easy to find. Just try.
  • »10.06.16 - 07:21
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 9285 from 2003/5/22
    From: Germany
    > It is easy to find. Just try.

    I did, and what I've found shows that Vampire/Apollo is way faster than 80 MHz 68060.
  • »10.06.16 - 07:59
    Profile
  • Order of the Butterfly
    Order of the Butterfly
    OlafSch
    Posts: 161 from 2011/11/16
    Quote:

    ppcamiga1 schrieb:
    It is easy to find. Just try.



    you completely miss the point...

    even if that would be true you are not able to get any 68060 80 Mhz accellerator out there or if only for a fortune

    if you talk about economic choices then your nick is also wrong because buying embedded hardware at price of high-powered standard hardware makes no sense either
  • »10.06.16 - 09:36
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 9285 from 2003/5/22
    From: Germany
    > embedded hardware

    It's not, as I told you yesterday.
  • »10.06.16 - 10:42
    Profile
  • Order of the Butterfly
    Order of the Butterfly
    OlafSch
    Posts: 161 from 2011/11/16
    Quote:

    Andreas_Wolf schrieb:
    > embedded hardware

    It's not, as I told you yesterday.


    yes it is a embedded system pretending to be desktop even beaten by any smartphone

    so even if it has some features typical for desktop hardware when you look at processor, graphics and so on it is embedded class

    perhaps we can agree it is a desktop mainboard with processors designed for embedded market

    2 Core with 1.8 Ghz is nothing I would today take seriously for new hardware anymore. New PPC processors (if developed at all) are mostly targetting embedded market or you have IBM with their series. Freescale is obviously dropping PPC now.



    [ Editiert durch OlafSch 10.06.2016 - 12:57 ]
  • »10.06.16 - 10:53
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 9285 from 2003/5/22
    From: Germany
    > it is a embedded system

    No, it's not. Please look up what the term means: https://en.wikipedia.org/wiki/Embedded_system

    > pretending to be desktop

    It does have features typical for desktop like RAM slot(s), SATA/USB/GbE, PCI/PCIe, audio (Tabor, Nemo) and HDMI (Tabor). It's not just pretending it has these.

    > when you look at processor, graphics and so on it is embedded class

    It's "just" the processor (CPU core) that is "embedded class". Graphics is provided by a desktop-class PCIe graphics card in a PEG slot. What does "and so on" mean?

    > perhaps we can agree it is a desktop mainboard with processors
    > designed for embedded market

    Yes, that's exactly what I've been saying all along.

    > Freescale is obviously dropping PPC now.

    There has been no Freescale for a while.

    http://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=11100&forum=3

    Yes, NXP is clearly in the process of dropping Power Architecture for new chips. According to their product longevity program, some Power Architecture chips will be available until 2029 (QorIQ) and 2031 (Qorivva), respectively.
  • »10.06.16 - 12:05
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4221 from 2009/1/28
    From: Delaware, USA
    13 years would be more than adequate for our purposes.
    I would still love to see an e6500 based system (and a T10xx mini itx board would also be nice).
    "Magnetic was troubled by my avatars and 'satanic' references" - Jim Igou

    "Never attribute to malice what can more readily explained by incompetence"
  • »10.06.16 - 12:47
    Profile
  • Acolyte of the Butterfly
    Acolyte of the Butterfly
    Posts: 129 from 2015/8/23
    Quote:

    Andreas_Wolf wrote:
    I did, and what I've found shows that Vampire/Apollo is way faster than 80 MHz 68060.



    I just check.
    Google still exists and returns many pages with screenshots from sysspeed on vampire.
    On all of them vampire is slower than 68060 80 MHz.
    You have to use google again.
  • »10.06.16 - 16:48
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 9285 from 2003/5/22
    From: Germany
    > Google [...] returns many pages with screenshots from sysspeed on vampire. On
    > all of them vampire is slower than 68060 80 MHz. You have to use google again.

    No matter how often I use Google, it will always return screenshots where Vampire2/Apollo is faster than 80 MHz 68060 in the benchmarks I asked you about.

    http://www.google.com/search?q=sysspeed+vampire&tbm=isch
  • »10.06.16 - 17:00
    Profile
  • Acolyte of the Butterfly
    Acolyte of the Butterfly
    Posts: 129 from 2015/8/23
    Google returns many pages with screenshots from sysspeed on vampire.
    On all of them vampire is slower than 68060 80 MHz.
    You have to use google again.
    Hint: use words sysspeed vampire and then check pages.
  • »11.06.16 - 15:09
    Profile