Slow defrag compared to copy

I'm trying to defrag a few 2GB files on a 1.3TB drive with half free, 6x500GB RAID10 Windows 2003 x64. The standard Windows defrag analysis report identifies a single 2G file in 3 segments. When I defrag that one file specifically with JKDefrag64Cmd, I only get a throughput rate in Perfmon of 3-5MB per second read and write, the current disk queue is always at 1, % idle is near zero, about 50 writes/second, and 64KB per write. It sees the largest gap as 130GB, and is moving 262144 clusters, but only one 64k block transfer at a time. It takes about 10 minutes to defrag each 2GB file.

If instead I just copy the file from D: to C:, I get a sustained 40MB per second read and write. The disk queue averages about 600, still 64KB per write, and it copies in about 45 seconds. If I then copy it back to D: (it's all in Windows x64 cache) it writes at 100MB per second for 20 seconds.

If the disks are given 10 blocks to move at a time, it can do them almost in parallel. But it appears the defrag is only asking to move one block at a time.

I understand you are using the standard Windows API. I see it supports "overlapped I/O" as the last argument to FSCTL_MOVE_FILE, and you're passing it NULL. Would that allow the API to asyncronously move blocks faster than one-at-a-time? Or is there a way to increase the number of bytes per write in a standard disk transfer?

Thanks, great product by the way.

Todd K