in reply to Sorting data that don't fit in memory
Here's a pointer.
If you can't fit the data into virtual memory, you're going to have to make a multi-pass sort, with intermediate results stored on disk. One algorithm you might consider is the "Radix" sort, which is a multipass, order preserving sort that is well-suited for very data sets with fixed-sized keys, and has O(n) behavior.
See Knuth, Vol. 1, Sorting and Searching for a complete description. As you read Knuth, consider how you'd use disk files for the intermediate partitions.
|
---|
In Section
Seekers of Perl Wisdom