SUMMARYTo evaluate the performance of a large‐scale digital system, the system can be modeled by stochastic timed Petri nets (STPN) and a performance index can be estimated based on the number of iterations in a Petri net simulation. When the performance evaluation is carried out in a large distributed environment, parallel calculation with incidence matrices for the STPN can make the Petri net simulation more efficient. In this paper, we propose an ordering method for the incidence matrices to reduce the communication costs in the parallel calculation of matrix operations. For this purpose, we first propose a method of blocking the incidence matrices based on the structure of the system by taking into consideration synchronous and asynchronous buses as structural elements. We also propose a method for converting submatrices of the incidence matrices into band matrices. This method is based on breadth‐first search in the STPN as represented by the submatrices. Finally, we demonstrate the benefits of ordering on the time required for a performance evaluation by carrying out a performance evaluation using the incidence matrices generated under the proposed method. In addition, we show that the ordering method can reduce the communication costs in a large distributed environment by estimating the utilization rate of calculation nodes.