in reply to The Upper Limit of Perl's Native Data Structures
Seeing as Graph::Directed is back by hashes, what is a realistic maximum size for the capacity of a hash or an array in Perl given factors such as the hashing function and local memory etc.
What do you mean by a "realistic maximum size"? The time to insert into a hash is fairly independent to the size of a hash, unless you are really unlucky and items hash to the same buckets. Every now and then, a single insert will take more time because the C-level array needs to be expanded and items rearranged, but this amortizes out over previous inserts. However, were you to plot a graph how long it takes to fill an N-element hash, you wouldn't get a fairly straight line, but you'd get a broken line. That's because at one moment, your program will begin to swap. But were swapping begins is highly dependent on your system.
As I insert this large graph into the Graph::Directed module I note what appears to be an exponential slowdown as I insert edges.I can't combine an exponential slowdown with Jarkko inserting 1.5 million edges. He isn't that old.
Abigail
|
|---|
| Replies are listed 'Best First'. | |
|---|---|
|
Re^2: The Upper Limit of Perl's Native Data Structures (abigail can be a punster too)
by Aristotle (Chancellor) on May 21, 2003 at 15:44 UTC | |
|
Re: The Upper Limit of Perl's Native Data Structures
by Abigail-II (Bishop) on May 22, 2003 at 22:59 UTC |