anthro@ut-ngp.UUCP (07/12/84)
Subject: Re: FORTH efficiency References: lanl-a.9623, <2867@ecsvax.UUCP> News Groups: net.lang.forth <> A further note: 'real' FORTH programs have very few constructs like the example A @ B @ + C ! FORTH applications make little use of variables, which are intended for global bookkeeping only. Keeping data flows on the stacks where they belong speeds up operations considerably. More reasonable/comparable source for the above example is + which accomplishes all that need be done in the usual case. FIG updates their literature rarely, but it was admitted some 8 years ago that someone had gotten carried away on the speed claim relative to assembly. It is noteworthy that once the names are disposed of, any moderatly long program will be shorter than a hand coded assembly version (try to recode some of the HL forth in-line sometime). This includes the runtime overhead. Mike Fischer anthro@ut-ngp