> I'm curious how much you gain in time by using the SMP SML/NJ compilation > stuff. I have never bothered partly because of fear that it would screw up. I didn't do any specific timings. It seemed a little bit faster, although watching top it didn't seem as though I was getting as much parallelism as I might have thought. I'll time make nj-mlton and make nj-mlton-dual from clean sources when I get a chance.