Intel X25-M SSD and File Fragmentation

Note (April 18 2009): OK, as a follow-up, I will have to remark a flaw in the below test. That is, it was done in respect to SSD-to-HDD copy, that means the HDD write speed bottleneck could be *well* below the (possible) file fragmentation speed penalty. So, I guess the test should be redone in a different way, that is copying from SSD to a faster media, i.e. RAM.

Since two weeks now I am the owner of an Intel X25-M SSD which fundamentally changed my T61p-s performance. From day 1 I was asking myself whether to defragment it or not... Technically I knew that memory is memory and spindle disks are spindle disks, so there was no apparent reason to have SSDs affected (on a practicaly approach) by file fragmentation, but... is it really so?

So, I decided to run a small test.

Test setup

Machine: Lenovo T61p, 4GB RAM, T7700 CPU @ 2.4 GHz

OS: Win 7 7068, NTFS

Primary HDD: Intel X25-M 80GB, Firmware updated to 8820 (which eliminates the sector remap fragmentation issue).

Secondary HDD: WD Scorpio Black (WD3200BEKT) 7200rpm 320GB.

The test subject was a single, 6.28 GB file (a vmdk VMware virtual disk file) which due to its size and NTFS compression was split in 4906 fragments.

The test itself

1. With the file as it was, i.e. quite heavily fragmented, copied it to a partition (with enough free space to prevent excessive destination fragmentation) on the WD secondary drive. Operation time: 123 seconds.
   
2. Defragmented the file (with sysinternals' contig utility). Result: perfect, i.e. file in 1 fragment.
   
3. Deleted the destination file on WD secondary drive.
   
4. Repeated the copy procedure. Operation time: 123 seconds.
   

Conclusion

File fragmentation effect on SSD read speed: nil, nada, nothing. 

What remains to be established in a subsequent test is free space fragmentation as this might affect write speed.