Rick,
Here is a test sequence about ":noname / : named " definitions....
Charles

\ _TEST FROM HERE_

EMPTY

: nsec ( -- n ) time&date 1999 - \ min second count from 1/1/1999
                 [ 12 31 * 24 * 3600 * ] literal * >r
                 [ 31 24 * 3600 * ] literal * r> + >r
                 [ 24 3600 * ] literal * r> + >r
                 3600 * rot + swap 60 * + r> + ;

0 value nsec-start

: [t nsec to nsec-start ; immediate
: t] nsec nsec-start - . ; immediate

variable va variable vb

: a 0 va ! va @ vb ! 1 vb ! vb @ va ! ;

: act a a a a a ;

: loop0 0 ?DO act LOOP ;

: loop1 0 ?DO i loop0 LOOP ;

0 value ntime

: evalcomputer
    2
    BEGIN
        nsec over loop1 nsec swap - 5 >
        IF to ntime exit THEN
        2*
    AGAIN ;

cr
..( Timing evaluation... wait max 15 sec... )
cr

..( time elapsed ) [t evalcomputer t]
cr

..( now, the same with some :noname... )

variable va variable vb

:noname 0 va ! va @ vb ! 1 vb ! vb @ va ! ; value ^a

:noname [ ^a compile, ^a compile, ^a compile, ^a compile, ^a compile, ] ; value ^act

:noname 0 ?DO [ ^act compile, ] LOOP ; value ^loop0

:noname 0 ?DO i [ ^loop0 compile, ] LOOP ; value ^loop1

[t ntime ^loop1 execute t] cr

cr .( on my pentium 300, result is near 11 / 7 ... )
cr .( in my class system wich align some word-xt table, result is 10 / 5 )
cr .( In all cases, the "noname" version is really faster ! )
cr .( NB: no ALIGN in this test )
cr
\ Last line here

.
Received on Tue Feb 16 1999 - 19:48:47 PST