Refine search for Storage

Intel X25-M SSD

Author	Edward Chester
Published	17th Sep 2008
Manufacturer	Intel
Price	£295.65 (Exc VAT)
as reviewed	£340.00 (Inc VAT)
Latest Price	Click here

Features
Performance
Value
Overall

5 comments

Email this to a friend

TrustedReviews Newsletters

However, there is an alternative type of flash memory, called Multiple Level Cell (MLC), that uses the same basic transistor but the number of electrons on the floating gate, and thus the threshold voltage, can be controlled more accurately. By cutting up the voltage range into more slices, more bits can be stored in each cell. In some cases this can be as many as 3 bits (8 voltage levels) but the most common is the 2 bit (four level) cell.

Thus, MLC technology is able to store more information per cell than SLC, resulting in a lower cost per megabyte and generally happy faces all round.

Unfortunately, there are a number of drawbacks to MLC memory.

First, the floating gate's ability to store electrons deteriorates each time the transistor is charged and discharged. Eventually this will lead to the threshold voltage no longer reaching its trigger point so the cell will never register as being charged and the cell will not store data properly.

For SLCs it takes around 100,000 charge/discharge cycles before errors can occur but because the gap in threshold voltages between each state in an MLC is much smaller, they have considerably less tolerance to deterioration. The result is MLCs can only last around 10,000 charge/discharge cycles before data corrupts.

Also, because of the more delicate way in which charging must be controlled, it takes longer to program an MLC, resulting in slow write speeds for MLC memory devices. It's worth noting, then, that Intel's E drives will use SLC memory and its M drives, MLC memory, which explains the performance figures stated on the previous page.

Having seen that each cell can potentially last only 10,000 cycles, I'm sure alarm bells are ringing for a few of you. However, you shouldn't be worried because by using some sophisticated data distribution and error checking algorithms, Intel ensures its drives can write 100GB of data everyday for five years before errors will occur, though they only warrant the drives for three years.

So, that's the basic SSD theory and it's largely the same for any of the SSDs out there but what's really interesting is what extra expertise Intel has brought to the party. Whereas most SSD manufacturers just make the memory and buy in a controller chip (or in the case of companies like OCZ, they buy in the whole lot, give it a tweak, and rebrand it), Intel has the expertise to design and make every aspect of its SSDs from the ground up.

The result is the near elimination of one of the most common problems with existing SSD solutions - the controller just can't keep up with the read and write commands it is dealing with. This scenario manifests itself as random pauses when doing seemingly innocuous tasks, which can make the experience of using an SSD little better than a conventional hard drive.

By using its own super-fast controller, Intel ensures that its SSDs will always exhibit that snappiness that is the biggest boon of SSDs in the first place.

Latest 4 of 5 Comments

Have your say: Leave a comment below about this article.

Peter Morris said on 17th September 2008

Thanks for the review, Ed. I work in professional audio and have ben testing fast USB drives as my audio disk for sometime with mixed results...

My need is for a drive that is always ready and has the fastest R/W possible so that multiple audio tracks can be read and written simultaneously. Simple enough and the fast HDDs do the job - up to a point. Typically a pro rig will have 1 disk for the system (including Logic, Pro Tools or other audio app), 1 disk for samples & virtual instruments, 1 disk for recording and playing back audio and 1 disk for video content (when scoring to film for example). A combination of internal disks and FW800 so as not to flood a single bus normally works well.

What concerns me is the inexplicable hits that SSDs sometimes take. I can understand it when an app has ben swapped out, when memory has ben paged out, when an HDD has spun down or encounters RPS latency. What causes the X25-M to take a bath on Windows shutdown? It's 60% slower than the Velociraptor. I've read other reviews that also have an inexplicable hit but in other areas. It's almost as if there is some kind of occasional initialisation issue or bottleneck.

It's hard to say to an artist: "Sorry - system fault. I lost that great take" so instabilities are far worse than predictable but inferior performance. There are strategies when the system behaves predictably but not when there are glitches.

Long preamble to a short question: are there glitches in SSD performance? Perhaps these are masked or averaged out in multiple tests but might manifest themselves according to Murphy's Law. What of the Windows Shutdown anomaly? Any ideas?

I'll still try it anyway, of course. When simultaneously reading and writing audio from the same disk (even with decent size buffers), the head movement is something I'd like to eliminate so that part of it is a no-brainer. As for capacity, the audio disk seldom needs to be bigger than 20GB for a recording session. I would archive after every session to RAID, whether using Velociraptor or SSD. 80GB is fine for work in progress.

Cheers!

Pete

Thanks for the review, Ed. I work in professional audio and have ben testing fast USB drives as my audio disk for sometime with mixed results...

My need is for a dr... more

HSC said on 17th September 2008

agree - need to be way cheaper
but only use I can see over a mechanical drive is in laptops, silent media pc's or if you need the ultimate performance....

... more

Simon Johns said on 17th September 2008

Is there any reason why the Intel drive was not matched against the latest OCZ Core 2 SSD? I appreciate that the OCZ drive uses the faster SLC chips (and clearly demonstrates just ... more

Ed said on 17th September 2008

@Simon

It wasn't a conscious decision to not compare the OCZ core range, we just haven't received one yet. From what I gather, though, the core drives are... more

See all 5 comments on this article.

Add your comment

You must be logged in to comment. Login or register here.