[lkml]   [1999]   [May]   [7]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
    SubjectRe: Overscheduling DOES happen with high web server load.
    Greg Lindahl writes:
    > > > Ok, you are right. The real problem is we are calculating goodness
    > > > O(A*B).
    > > >
    > > > A= Number of processes on the runqueue
    > > > B= Number of times schedule is called
    > >
    > > Don't underestimate the cost of searching the run queue. Put 10
    > > processes on the run queue and you more than double the cost of
    > > schedule() to another process. For 90+ tasks on the run queue, the
    > > cost is dominated by the linear search.
    > You are violently agreeing with Phil. The linear cost of searching the
    > run queue is represented by A in his expression. That was the point of
    > my little hack, which was to always take the first schedulable process
    > in the queue, merely to see if that made life much better. And it
    > seems to.

    Maybe I didn't express myself clearly. I thought his (or someone's)
    point was the cost of goodness() calculations dominated, the
    implication being that reducing the cost of goodness() would speed
    things up a lot. While true, the cost of scanning the task structures
    is also high. But in essence, I agree with Phil.

    Someone else suggested that the cost was in all those context
    switches, which I dispute. Switching state is relatively cheap. And
    this was my point (and as you say, Phil's too).



    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to
    Please read the FAQ at

     \ /
      Last update: 2005-03-22 13:51    [W:0.020 / U:76.824 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site