| 摘要 |
A technique for searching in a source sequence for a target sequence using parallel processing. Each symbol of the target sequence to a corresponding processing element, where the processing elements are connected to form an array and at least one processing element supplies an accumulated score to a succeeding processing element in the array; supplying a symbol from the source sequence in parallel to all of the processing elements, so that each processing element receives the same source sequence symbol; determining a score based on the source sequence symbol and the processing element's target sequence symbol for each processing element; adding the score for a processing element to the accumulated score received by that processing element from a preceding processing element to generate a local score; comparing the local score to an accumulation value; when the local score is less than the accumulation value, supplying a default accumulation score to the succeeding processing element as the accumulated score; and when the local score is not less than the accumulation value, supplying the local score to the succeeding processing element as the accumulated score.
|
