re new SML/NJ

[Henry Cejtin]

I am not really arguing that we should switch to MLRisc, since as you say, it
would be very hard because of  the  non-standard  aspects  and  very  painful
because  of  the  bugs  that are almost certainly there.  The point is though
that I am curious how much better, if any, their local-ish code generator is,
and  how  much  that is buying them.  If we knew it was a lot, it would be an
argument for us to fix ours up, while if they are getting squat, then we  can
continue without worrying too much about it.

Speaking  of  which,  I  tried  the silly floating point test that appeard in
comp.lang.functional
    fun test (n, x) =
           if n = 0
              then x
              else test (n - 1, x + Math.cos x)
and our code was definitely nicely faster than gcc:
    gcc     278.9 nanoseconds
    MLton   234.2 nanoseconds
so we are 20% faster.  I looked at the code, and ours still does one floating
point  load  and  one store per loop, but the C version is doming some really
funny stuff.

In the best of all worlds, the back end would figure out  that  the  floating
point register only has to be stored when the loop finishes.