No subject

[Henry Cejtin]

Well  it  looks as if they must have improved the I/O speed by quite a bit in
SML/NJ.  I'm still running 110.9.1, but a  simple  version  of  wc  does  the
following:

        Native C: .04 seconds
        MLton:    .34 seconds
        SML/NJ:   .48 seconds

Mind  you,  I  did  use  the  TextIO.input  procedure.   When  I  switched to
TextIO.input1 the times became:

        SML/NJ:  1.15 seconds
        MLton:    .25 seconds

I guess that SML/NJ sped up because it didn't have to  allocate  the  string?
Seems a bit weird.

Anyway, the size is much better for MLton.

        Native C:         7,484 bytes
        MLton input:     32,264 bytes
        MLton input1:    30,152 bytes
        SML/NJ input:   349,968 bytes
        SML/NJ input1:  348,944 bytes