I agree with this. There has been a few posts here in the past, where it was shown that the OS optimizes the processing of files, and it dosn't do much good to split the file and process the chunks in different threads of the same process. Disk IO will be the bottleneck. Different machines is a good idea. Or maybe if you were on a fast scsi system, and could put the different chunks on different scsi disks it would do some good.