Announcement

Collapse
No announcement yet.

Old Motorola 68000 Systems Can Finally Move Away From Linux's Deprecated IDE Code

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • oiaohm
    replied
    Originally posted by gamerk2 View Post
    It's worth noting the 68k derived architectures are still used as microcontrollers due to having quite good power/performance for the job they need to do.
    Its not just that. There are quite few different industry controllers based around the 68k their code base is well and truly validated. Starting on a new microcontroller would mean having to deal with the differences all over again. With the fact that fgpa version of the 68k exist that have exactly the same performance as original asic versions there is no need for those to update any time soon. The fun reality as what the Apollo 68080 shows modern fpga can make a current day 68000... faster than the originals so its not that hard to limit performance to generate identical performance to the no longer made chips.

    68k in modern fgpa are not the highest power effective option but they are more power effective than than the original asic chips as well. Not all jobs do you need perfect power/performance there are quite a few 68k existing usage where the most important thing is that you don't kill anyone due to a malfunction those areas the improved power to performance gains are just extras on top.

    Leave a comment:


  • gamerk2
    replied
    It's worth noting the 68k derived architectures are still used as microcontrollers due to having quite good power/performance for the job they need to do.

    Leave a comment:


  • oiaohm
    replied
    Originally posted by MadeUpName View Post
    I don't get this. The kernel guys want to EOL my 4 year old Celleron HTPC because it doesn't have some CPU extension but an Apple II that definitely doesn't have it is fine?
    There is an answer to this and not the Apple II.

    Originally posted by schmidtbag View Post
    I assume this has something to do with the fact m68k has been long dead whereas x86 continues to evolve. Backward compatibility gets increasingly complicated as newer technology comes about, especially if that newer tech either depends on other tech, or, if developers become too reliant on it.
    This is a mistake the m86k are still on the market as FPGA chips. 4 year old Celleron where are you going to get that chip new. The reality is m86k has a FPGA form that perform absolutely identical to perfectly functional old m86k.

    http://www.apollo-core.com/index.htm?page=features
    There are also new 86000 instruction set cpus being made that FPGA base some of these support a 64 bit instruction set. Yes the example here is not Motorola and used to make AMIGA clone machine and used as a upgrade to old AMIGA computers to make them run many times faster than they use to by replacing the old Motorola cpu.

    The difference here is the reality is you can get new 86000 instruction set cpus that run in FPGA and the old Celleron chip buying new can be a very hard problem. Please note being FPGA implementation with the 68000 is really simple to upload a different FPGA bytecode to make a version of the 86000 missing instructions so its really easy to replicate old 68000 cpus with modern FPGA chips on brand new hardware.

    Its a real big mistake to think that m68k is a dead platform. M68K is a platform alive and evolving as a FPGA implemented CPU. M68K only really stopped developing as a ASIC solution for the past 20 years it has been developing as a FPGA solution.

    The problem with supporting old GPU and old CPU is getting new defect free ones to test software against. The reality is 68000 does not have that problem due to all the FPGA implementations that are defect free to test software against. Yes the M68K is not the only cpu alive and well as FPGA implementations that the Linux kernel still supports. Really old and still support by the Linux kernel more often than not is a case that that bit of old hardware can be implemented perfectly in modern day FPGA chips.

    Please note a M68k in asic was only a "68,000" transistor solution your a 4 year old Celleron x86 is in your billion of transistor class cpu. So a 4 year old x86 chip is too complex to implement at this stage in a FPGA yes even the biggest FPGA on the market does not have the required space to implement a modern x86 cpu.

    Now 16 bit and 32 bit x86 FPGA implementations do exist just nothing that is a 64 chip or with all the 32 bit added instructions that the amd and intel 64 bit cpus added to the 32 bit instruction set.

    Leave a comment:


  • set135
    replied
    Originally posted by eltomito View Post
    Shit! And somebody said Linux would deprecate Sandy-Bridge (2012) era hardware because it was TOO OLD! Well, fuck my blitter and move.w #f0c4, a0!
    Well, there are distributions, and there is the kernel. The kernel still supports i486 I think. Distributions have been dropping olde x86 support for a while; some only bother with amd64 now. You can still find some that claim i686, but that will probably not include a lot of i686 cpus, because they tend to require the CMOV instruction, and maybe some others. But yeah, some distributions may start requiring functionality that older cpus lack. When the kernel itself deprecates something, it tends to be ancient hardware that no one has or no one cares about maintaining anymore.

    Leave a comment:


  • leech
    replied
    Originally posted by thunderbird32 View Post
    Based on the driver name, I assume this is based off an IDE driver written for the Atari Falcon? Humorous that a system as rare as the Falcon should be the basis for supporting machines as common (comparatively) as the Macintosh.
    I always thought it was the opposite, that the Falcon driver for MiNT was based upon the libata driver from Linux.

    Leave a comment:


  • eltomito
    replied
    Shit! And somebody said Linux would deprecate Sandy-Bridge (2012) era hardware because it was TOO OLD! Well, fuck my blitter and move.w #f0c4, a0!

    Leave a comment:


  • milkylainen
    replied
    Originally posted by schmidtbag View Post
    muncrief
    Often when academics, researchers, and scientists make fun of an idea, it either shows a lack of understanding or it shows they don't want to be proven wrong.
    Second that. The proven wrong sits so badly with some individuals, they'd rather end the world than be proven wrong.
    Ergo, they'll do anything to keep and validate their view, status quo etc.
    You'll find them everywhere. Even in medicine, where your life would depend on someone willing to accept that they were wrong and do something about it.

    Leave a comment:


  • milkylainen
    replied
    Originally posted by muncrief View Post
    I created a biological neural simulator in my early 20s on a 68000 based Atari 1040ST. I invented four unique software structures - reception, conduction, integration, and transmission - to create dendrites, somas (cell bodies), axons, and synapses for a variety of neurons. And these were true to life neurons with everything from passive, electrically, and chemically gated ion channels to anterograde and retrograde transport mechanisms.
    I had a similar clash with the "institutionalized" when I was about to pursue a PhD. That was almost 25 years ago.
    This was before GPU acceleration was a thing. I saw the potential very early on and wanted to extend into application research for numerical acceleration on future hardware. Was completely downplayed as "We already have large CPUs and servers..."
    Luckily the world went that way anyway. I was so pissed about the limited and obtuse world view that I've never set another foot in university again.

    Leave a comment:


  • schmidtbag
    replied
    Originally posted by MadeUpName View Post
    I don't get this. The kernel guys want to EOL my 4 year old Celleron HTPC because it doesn't have some CPU extension but an Apple II that definitely doesn't have it is fine?
    I assume this has something to do with the fact m68k has been long dead whereas x86 continues to evolve. Backward compatibility gets increasingly complicated as newer technology comes about, especially if that newer tech either depends on other tech, or, if developers become too reliant on it.

    Leave a comment:


  • MadeUpName
    replied
    I don't get this. The kernel guys want to EOL my 4 year old Celleron HTPC because it doesn't have some CPU extension but an Apple II that definitely doesn't have it is fine?

    Leave a comment:

Working...
X