Z.
uh huh... But most people who post here couldn't really read the stuff. I routinely 'tone down' everything I post. The idea is to communicate, not obfuscate. 90% of what I do is in ANSI C.
Parksie -
don't you try to inline, jam, etc, your core code? ie., optimize the living doo-doo out of your code? Especially the core code - primitives? We clock() on everything because the stuff gets called hundreds of millions of times a day. Literally. 5% gets you five minutes.
We have tables that have 100 million+ records. Full table scans net lots of traffic into the core modules. For example, Archive-purge runs over 10 hours. We want it to run in 5 or less.
It cleans out datasets.
I know what you mean. Things that seem really obvious to me now wont even occur to someone who has only been using c/c++ for a few months... and again, in a year or so, there will be things that are just natural where they wouldnt even have occured to me now.
Originally posted by jim mcnamara Parksie -
don't you try to inline, jam, etc, your core code? ie., optimize the living doo-doo out of your code? Especially the core code - primitives? We clock() on everything because the stuff gets called hundreds of millions of times a day. Literally. 5% gets you five minutes.
Yeah, I do. We have to do about 100 calculations per finite element, of which some meshes have a few hundred thousand.
I did some *really* horrible-looking parser code a few days ago, but it sped it up from 6 hours to about 40 seconds.
There's nothing wrong with optimising. Just don't do it too early. Get it working first!
I refuse to tie my hands behind my back and hear somebody say "Bend Over, Boy, Because You Have It Coming To You". -- Linus Torvalds