In recent years a large number of evolutionary and other population based heuristics were proposed in the literature for solving NP-hard optimization problems. In 2015 we presented a Discrete Bacterial Memetic Evolutionary Algorithm (DBMEA) for The Traveling Salesman Problem. It provided results tested on series of TSP problems. In this paper we present an improved version of the DBMEA algorithm, where the local search is accelerated, which is the most time consuming part of the original DBMEA algorithm. This modification led to a significant improvement, the runtime of the improved DBMEA was 5– 20 times shorter than the original DBMEA algorithm. Our DBMEA algorithms calculate real value costs better than integer ones, so we modified the Concorde algorithm be comparable with our results. The improved DBMEA was tested on several TSPLIB benchmark problems and other VLSI benchmark problems and the following values were compared: - optima found by the improved DBMEA heuristic and by the modified Concorde algorithm with real cost values - runtimes of original DBMEA, improved DBMEA and modified Concorde algorithm. Based on the test results we suggest the use of the improved DBMEA heuristic for the more efficient solution of TSP problems.