Morphos 3.19/3,20 ?
  • Moderator
    Kronos
    Posts: 2306 from 2003/2/24
    @Cego

    Unless you want MorphOS on your phone, smart TV or fridge, desktop is what it is.

    You also don‘t need 500W to get faster single threaded than anything ARM.

    And thats the issue, as it stands today we do need single threaded as fast as possible and even with the best Intel or AMD it will be the limiting factor.
  • »19.09.24 - 14:27
    Profile
  • Priest of the Order of the Butterfly
    Priest of the Order of the Butterfly
    ernsteiswuerfel
    Posts: 553 from 2015/6/18
    From: Funeralopolis
    My 2 cents so far:

    These 'MorphOS on alternative arch'-discussions pop up more often when the last released MorpOS version is already more than a year behind. So maybe even a new PPC-MorphOS version will silence them (for some time). ;-)

    While valid points are nicely discussed I doubt a discussion in an user thread will convince one or more devs to scrap their already existing x86_64 MorphOS Alpha and start on another architecture from scratch. Also this would postpone 'MorphOS on alternative arch' even more... While I personally also would 'like' an ARM version, the x86_64 route already has been taken so x86_64 it will be.

    What hasn't been broadly discussed is the documentaion available/needed to port for ARM boards. These boards seem to run 'nice' on Linux, but it is easily forgotten that most of the time they only do with vendor kernels and binary blobs (especially for GPU drivers). So practically most of the time you would port to a board running an outdated vendor kernel + proprietary drivers and with scarce documentation. The Raspberries would be among the best supported, but still RPi 5 has no full mainstream kernel support (might probably take another year) yet... RPi 4 would be ok kernel-wise, but I don't know about the GPU.

    So in the ARM case a big amount of reverse engineering would be needed to get things going, as in the Linux world. On the x86_64 side this situation is much better. There are also a lot more standards used (UEFI, ACPI, ...) An AMD B450 board is not totally different from another AMD B450 board. An ARM-SBC running a Cortex-A76 CPU probably is totally different from another ARM-SBC running a Cortex-A76 + documentation is scarce...

    Also as Kronos states for MorphOS in it's current state only single core performance would be of interest. In this regard the RPi 3 most certainly is slower (1,4 GHz CPU A53 vs. 2,3 GHz G5), also RPi 3 storage via USB is slower than SATA-1. Only the RPi 4 might get en-par performance (1,8 GHz A72 + larger cache than G5) or is slightly faster. USB-performance also is better than RPi 3.

    Regarding the efforts needed to get MorphOS to work even on one ARM board, which is not magnitudes faster than currently supported PPC hardware, the x86_64 route still seems a good choice for the time being. Especially as quite some work has already been done.
    Talos II. [Gentoo Linux] | PMac G5 11,2. PMac G4 3,6. PBook G4 5,8. [MorphOS 3.18 / Gentoo Linux] | Vampire V4 SA [ApolloOS / Amiga OS 3.2.2]
  • »19.09.24 - 17:46
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12132 from 2003/5/22
    From: Germany
    > Raspberry Pi 3B+ already does as well as a PowerMac G5.

    Like ernsteiswuerfel, I really doubt that for per-core performance.
  • »19.09.24 - 18:24
    Profile
  • Acolyte of the Butterfly
    Acolyte of the Butterfly
    Georg
    Posts: 110 from 2004/4/7
    Quote:

    ernsteiswuerfel wrote:

    Regarding the efforts needed to get MorphOS to work even on one ARM board


    Run the A/Box hosted or virtualized like is already the case on PPC. But instead of Quark use something which already works on the board. Hide it and don't tell anybody and everyone will think they are running native. Like they do on PPC.
  • »19.09.24 - 18:46
    Profile