AN APPROACH TO DERIVE TIME COSTS OF SEQUENTIAL COMPUTATIONS

Before deriving the time cost of a sequential computation, current time-cost analysis techniques need to perform a flow balance operation in order to determine the number of times each node in the computation is executed. This paper presents a new time-cost analysis technique. The new technique does not perform the flow balance operation. In deriving the time cost, the new approach deletes all of the dependent flows and reduces the computation to a single node graph. During the reduction process, the time cost and the flow count associated with the dependent flow to be deleted are properly distributed within the computation. It has been found that this technique is more efficient than traditional approaches. © 1990.