lkml.org 
[lkml]   [1997]   [Jun]   [11]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    Date
    From
    SubjectRe: "obsolete" hardware
    Hi,

    > Of course there is another way of looking at this. There are lots
    > of current and upcoming applications that do require lots of horsepower.
    > By the same token there are fewer and fewer (if any) applications being
    > written that target a 486 (let alone a 386) as a minimum requirement. Try
    > running anything X/Motif based and you'll see what I mean. If Linux
    > developers concentrate their attention on trying to be compatible with
    > obsolete hardware as opposed to taking advantage of cutting edge features
    > other operating systems will overtake Linux by running slower on faster
    > hardware - not exactly a pretty picture.
    >
    > Bottom line is I'd rather say "Linux runs faster on a quad P7" than
    > "Linux still runs on my 2mb 386" a year from now. A well executed design
    > means nothing unless it performs. Nothing PERFORMS on a 386. Lets not
    > loose sight of our goals here for the sake of keeping backward
    > compatibility.
    >
    > Take as a warning the longevity of the PC bus architecture. Would you
    > care to guess why we're still stuck with it? Backward compatibility. Its a
    > dirty word.

    (1) The first problem is that if you want to remove dirty hacks for broken
    hardware by removing support for the _old_ hardware, you won't win as lots of
    new hardware are broken as well.

    (2) Linux is certainly meant to offer much more than running CPU-eating
    X applications. Those other capabilities (including network routing and
    bridging) require only little CPU power and a quad P7 won't noticeably
    improve their performance, so why buy it instead of using an old 386 and
    saving $6000. Well written applications seldom need lots of CPU time.

    (3) We were talking about things like FPU emulation which don't need much
    time to maintain -- just to keep them consistent with the rest of the kernel.
    And using FP in the kernel is of a small value, it would also mean saving
    and restoring them on every syscall or take care of saving/restoring at every
    use.

    => The answer is "No."
    Have a nice fortnight
    Martin

    \
     
     \ /
      Last update: 2005-03-22 13:39    [W:0.039 / U:121.288 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site