| 摘要 |
The invention comprises a method of completely sorting an unsorted data table in a single pass through the data. The method uses a relatively large amount of computer memory, but sorts the data at great speed. Specifically, each unsorted data word is scaled to a size equal to or less than the number of address locations in a sort table. The scaled value created for each unsorted data word is converted to an address increment which is added to the initial address of the sort table. The unsorted data word is then stored into the sort table at the above calculated address that is related to the value of the data word. In this way, most words are placed in a sorted arrangement without comparison or iteration. The invention also comprises a method for resolving conflicts where a calculated address for an unsorted data word already contains a sorted data word. In this case both words are then sorted with respect to each other to create a data subclass and stored in an auxiliary storage table. In the case of subsequent conflicts, conflicting data is sorted and then moved en masse to yet a further location in auxiliary storage. A further element is a method of gathering and merging the data which has been sorted as described above. The gathering mode eliminates the null values between sorted data words and places the subclasses of resolved conflicts in sequence in the data table.
|
