hm, for quite a long time i was wondering if this construction...
for (loops) { n = some_pointer->some_member.some_number; }
...would be (noticeably) slower than this construction:
my_n = some_pointer->some_member.some_number;
for (loops) { n = my_n; }
while one would probably assume it's obvious the latter version is faster, someone somewhere suggested it wouldn't make a (noticeable) difference due to compiler optimization.
now I've done a few test runs (pretty much like the pseudo code above), and I think it is indeed faster. results vary strongly, esp. on my 8-core Linux machine, but on the Amiga the latter version of the code is near-consistently about 25% faster than the former. That's certainly well noticeable. I bet there are a hundred things to keep in mind, e.g. loop-size vs. CPU cache, size of struct, location of member, ... I don't know... so I don't trust my results.
what do you think? any experience, opinion?
btw., the reason behind my research is that when creating a program I have a tendency to create e.g. a large struct that holds my program's data, settings, pointers, etc.. I'm wondering if constantly accessing data via such a struct is really bad practice, performance-wise. ;-)
Please wait...
0 files
