[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

GC speed



    Date: Wed, 27 Apr 88 14:23:36 CDT
    From: wu at cs.wisc.edu (Felix S.-T. Wu)
    To:   t-discussion at mc.lcs.mit.edu
    Re:   GC speed

    I am running T3.1 on a Vax8550 (~10 MIPS), the garbage collection
    seems to be a little slow to me.  For instance, it took about 4.6
    seconds (timed by the provided "time" form) to GC a just started T
    process (with 4M bytes for each half of the heap).  What I would
    like to know is: Is this GC speed kind of slow or normal (or fast) ?

I believe it is slow compared to other lisp implementations.
MIT's C Scheme is alleged to GC in a fraction of a second even with
a nonempty heap.  A few seconds is normal, I think, for Lucid and VAX LISP.

    If it is slow, why is it?

Because there isn't enough manpower to do everything that needs to be
done.  It was written to be correct first, fast second.  It just hasn't
been tuned very much.

    Is it inherent because of T's data representation or what ?

I don't think so, but the complexity of the representations doesn't help.
I think there may be some hacks in the next  release to speed it up.