home *** CD-ROM | disk | FTP | other *** search
- On 08-Apr-97, Rune J. Keller sat on a keyboard and produced this mess:
-
- > This very good file explains different ways to improve speed in your
- > AMOS-programs.
-
- Thanks!
-
- > However some of the tips really surprised me. Ben states that A=A+1 should be
- > faster than both Inc A and Add A,1. He also writes that Repeat ... Inc ...
- > Until should be faster than For ... Next. I have been doing some tests on my
- > system and it clearly proves that this is WRONG!
-
- Whoops!
-
- > I can not believe that Ben Wyatt has not checked his tips, so I must conclude
- > that the difference in speed is different on different system (I do not think
- > that Ben has a 68060?).
-
- You're right. I tested most of the routines on an 020 and some on an 030 and
- they were faster on them. I think these are probably more average systems.
-
- > Anyway I am working on a small program that will test the speed of different
- > commands, and I am thinking that it should be run on different systems, so we
- > could clear out which commands is fastest on which system. The rest of the
- > message is an output of my not-yet-finished speed tester.
-
- I'd like a copy of this program when you finish it. I'll try it on my system.
- Does the program take CPU caches into account? That causes some wierd results
- on my 030.
-
- > Do anyone has some comments on this?
-
- Sounds good to me. I'd like to see some results for the > and >= comparision
- tests. I can't believe the one is slower than the other.
-
- > [SNIP]
-
- I'll add these test results to the list.
-
- Bye ____________________________________________________
- / \
- / Ben Wyatt - bwyatt@paston.co.uk or b.wyatt@uea.ac.uk \
- \ http://www.paston.co.uk/users/bwyatt/index.html /
- \____________________________________________________/
- (c)1995-97 Very Interesting Signatures Ltd.
-
-
-