12 Replies to “A BSD without i386”

  1. This is not really comparable to what John proposed (dropping i386 support from DragonFly base). PC-BSD is technically “just” a distro of FreeBSD and I’ve heard nothing of FreeBSD planning to drop i386 support from their base.

    PC-BSD isn’t released for a lot more architectures which FreeBSD proper supports.

  2. It’s not the same, but it’s a blurry line between the two. The fact that it supports a limited number of platforms already makes it not surprising that it’s the first to go… but it is the first.

  3. Solaris 11 (hey it has at least some distant BSD ancestors) dropped support for i386 as well ;)

  4. Is not the only significant difference between 32bit and 64bit processors just the limitation of the former to 4GB of main memory? I wonder why the architecture matters at all. Shouldn’t software written cleanly and in a high level language be unconcerned with the machine that is compiled for? A clean software design would also not prevent optimization for specific machines either.

  5. Supporting 2 architectures takes more time because you need to generates more ISO, build more packages, verify that it builds correctly on both archs.

    I think PC-BSD should drop i386 support as it aims computers that can run ZFS (which shouldn’t be used on 32 bits systems !). Why would you run PC-BSD if your computer is old and doesn’t have a lot of ram ? Just install FreeBSD, you won’t use firefox or libreoffice anyway as it takes too much memory/cpu.

  6. Computers without “a lot of RAM and CPU horsepower” need not necessarily be old computers. You might not yet be able to find a fanless 4x4x1 inch sized computer that is cheap and 64bit. That’s why some folks might still see use for 32bit x86 support for years to come.

  7. @skdj : However, the last i386 release of DragonFly will also work for years to come. I ran Solaris 10 for 6 years without touching it. Ditto for FreeBSD 6, although one box had a bad drive so I upgraded to FreeBSD 7.2 and ran that for 4 years.

    Once installed, there’s usually not a driving reason to upgrade the OS itself. None of the pro-i386 folks has answered why long term support on the last i386 is not satisfactory, or why this old/weak boxes require the latest release when the 3.x releases are so robust now.

    The EOL of a platform doesn’t mean users of that platform are cut off from DragonFly, only the very latest releases.

  8. I think it’s OK to drop support for i386. however, there are still some computers out there, which are quite new and though are still i386 architecture, like my acer aspire one netbook with an atom processor. Linux distros are making a fine step in between like they do not support anything but i686 with PAE extension. that’s something one can live with for a few years and would still support old hw like AMD AthlonXP can be often found among old computers that are still in use.

  9. It would be a pity to drop the i386/32-bit as people still use it (check the FreeBSD questions list). Having a 4Gb RAM address limit is meaningless if you don’t have that much RAM, and 32-bit code is more compact. Likewise, the 64-bit sector threshold that stops gmirror working on drives beyond 2Tb (512-byte/sector) is the only reason many will have to go to ZFS right now.

    In some parts of the world, 32-bit is the only option and I think it’d be a shame for FreeBSD to freeze them out. I don’t really want to have to direct people to Ubuntu as the only option.

    For the sake of balance in this comment, I will concede that 32-bit is really dead. Microsoft says so.

Comments are closed.