Vector processor for processing database without special tags for vector elements

DETAILED DESCRIPTION It is an object of the present invention to provide a vector processor which does not need to add tag information and has a high processing efficiency.
The above object is achieved by providing means for generating information which enables identification of each vector element in a vector operand such as vector element numbers in an operation unit or main memory addresses of vector elements and storing such information in the main memory as tags for the operation result.
A part number of a part name table 701 is stored in a first input operand X and a part number of a price table 702 is stored in a second input operand Y, and the element of X and the element of Y are compared.
If both elements are equal, a pair of operand counters of the input operands are generated as tag information to identify both elements and they are outputted as elements of a third output operand Z.
The operands X and Y and the counters i and j are initially "1" and the comparison of X(1) and Y(1) is equal (P2=P2).
Accordingly, the pair (i, j) or (1, 1) is stored as Z(1).
Then, the operand counters i and j are incremented by one and X(2) and Y(2) are compared.
Since they are not equal (P3<P5), the pair (2, 2) is not stored and only the smaller operand counter i is incremented by one, and similar operation is repeated.
When the operand counters i and j are 3 and 4, respectively, they are gain equal (P8=P8) and the pair (3, 4) is stored as (Z2).
The output vector Z shows the lines of two tables to be joined which have the same part number.
A join table may be prepared based on the vector Z.
Since the tag information for identifying the line is derived from the operand counter which indicates the progress status of the vector operation, there is no need to previously add the tag information to the vectors X and Y.