Re: [Matlisp-users] m.*! not native?

[Raymond T. <to...@rt...>]

>>>>> "Jefferson" == Jefferson Provost <jp...@cs...> writes:

    Jefferson> On 1/31/02 5:40 PM, "Tunc Simsek" <si...@ee...> wrote:
    >> what do you mean by native.  both allegro and cmucl produce native code.
    >> looking at mtimes.lisp; it looks like all the declarations are in place
    >> for m.*! so the only reason that it may be slow is if the optimization
    >> flags were not set during compilation.  I personally never checked that
    >> they are.  It is worth checking this.

    Jefferson> Sorry, I misspoke: by native I meant to say implemented in fortran.

    Jefferson> Here are the average times of a few operations.

    Jefferson> operation  | msecs  (mean of 100 runs)
    Jefferson> -----------+---------
    Jefferson> (m.*! a b) | 1351.6
    Jefferson> (m.+! a b) |    8.6
    Jefferson> (m* c d)   |   15.2

    Jefferson> A,B = 500x500 real-matrices
    Jefferson> C = 500x1 real matrix
    Jefferson> D = 1x500 real matrix

You do realize that a*b is O(500^3), a+b is O(500^2) and c*d is also
O(500^2).  So it's not surprising that a*b takes 157 times longer than
a+b and 88 times longer than c*d.

To speed things up, you may want to get a copy of ATLAS and build
matlisp with that if you haven't already.

Ray