> Are you sure that the hash resizing was actually triggered in your tests? I tested it with a 1Gb+ executable, and the change made a significant improvement, and the resize was definitely happening.