What is disk fragmentation ?
When a file is written to disk a block is allocated. As more and more is written to the file other blocks will be allocated. Ideally all the blocks of the file would be next to each other in one large chunk. This means that when the file is read the disk head doesn't have to move. It can "scoop" up the file in one go.
Fragmentation occurs when the allocated blocks are scattered across the disk.
One of the speed benefits come from the fact that PFS-2 maps a file with a-nodes. This carries the details of one chunk or fragment. Ideally a file will be mapped by one a-node. When you "seek" into a file PFS-2 can do this quickly. FFS has to look at the file header and extension blocks.