You do know that in those benchmarks a big chunk of zeroes (or random data) generated in memory is written to the HD being tested, right? It's done like this so that the "source" of data isn't a bottleneck for testing the destination's write speed, since memory is much faster than any HD.
ImgBurn, on the other hand, reads data from the source and writes to the destination as fast as your system can handle, and since you're reading and writing on the same volume, that's why it slows down a lot.