Intel Responds to Fragmentation with New X25-M Firmware
Firmware Details, System Setup, and Pre-flash Setup
Firmware 8820 Details
Quoteth Intel: "It should be noted that any of our SSDs will see periods of reduced performance after significant random write fragmentation (white noise random fragmentation, not what Windows generally does with it’s “random” writes) as the drive cleans this all up mixed with additional data being written. This new firmware does not change this fact. What it does do is prevent the drive from getting into a state where further sequential writing will not recover the drive. You should see this with HDTach, or a large file copy, or just general use, etc. So, if a drive is in what previously seemed to be a permanently degraded state (as discussed, we still feel this is highly unlikely for a client PC user), and a user installs the new firmware they will feel an instant improvement for any sequential operations, which will get better in time as the drive cleans itself up further. This firmware will also prevent the user from getting into such a drastic state of fragmentation, and generally help ensure the sequential write performance is as good as it can be at any moment. This change really has no significant impact on random performance."
Hmm, so this firmware should in theory be able to dig a severely fragmented X25 out of it’s hole, and with such a mechanism in place, it should also more aggressively ‘hold the line’ as far as sticking to a write speed of 80 MB/sec. Yummy.
Test System Setup
Nothing special about the setup for this one.
|Hard Drive Test System Setup|
|CPU||Intel Core i7 920|
|Memory||Corsair Dominator 6GB DDR3-1600|
|Hard Drive||Intel X25-M 80GB SSD|
|Sound Card||Asus Xonar D1|
|Video Card||EVGA Geforce GTX 260 Core 216|
|Video Drivers||Geforce 181.22|
|Power Supply||Corsair 620HX|
|Operating System||Windows XP X64 SP2|
In early March, after beating the crap out of my own personal X25-M to collect data for my first article, I wiped it and imaged my OS back onto it, returning it to normal use.
It is important to note that I *do not* image the OS back with a sector-by-sector copy. Doing so is not representative of any kind of typical user action. The vast majority of users are going to do a ‘real’ OS install on their shiny new SSD. OS installs, at a minimum, involve copying individual (small) files to the drive, which puts the X25 into write combining mode. This causes fragmentation that *will* effect performance. Tools like Paragon Partition Manager that enable migration of a live OS over to an X25 will perform a file based copy. This is because copying a live windows partition is much more practical when done at the file level as opposed to a sector-by-sector copy. A file-based imaging will still be ‘easier’ on a drive than a full windows install, as even more small writes take place while the windows installer builds the registry and saves various settings.
Over the following weeks / months I used my system normally with one exception: I avoided using my X25 for temporary storage / manipulation of very small files. This is what took my drive over the edge last time. Aside from that one adjustment to my usage pattern, I did my normal thing. I evaluated a newer build of Windows 7 on a second partition, built a few VM’s on the primary partition, checked email, surfed, and threw in some occasional gaming. I don’t subscribe to the drastic measures employed by some SSD enthusiasts. In my opinion, most users are not willing to go to these extremes, and they shouldn’t have to so long as the SSD is properly executed in the first place.