cancel
Showing results for 
Search instead for 
Did you mean: 

Intel X25-M *G1* extremely high Disk Queue Length

idata
Esteemed Contributor III

I purchased an Intel X25-M (G1) 80GB SSD in October 2009 when I built my computer - so a little over a year ago. Initially it was beautiful, very fast and reliable. Expensive, at $240 from Newegg, but I figured I was paying for quick boot times, snappy OS, fast core applications and web browsing.

Unfortunately in the last 3-4 months, performance has been degrading gradually, frmo noticeable, to bad, to outright terrible. It seems to manifest most often as insanely high disk queues in Windows 7 Resource Monitor. When I look in the Disk tab, I see the following, without fail, every time my system does its periodic lock-ups:

http://i163.photobucket.com/albums/t299/anon09876/discqueue.png Example 1

http://i163.photobucket.com/albums/t299/anon09876/discqueue2.png Example 2

Both shots are literally from the last few minutes as I was typing this message. Response time column in Resource Monitor skyrockets (sometimes 6,000+), Disk Queue Length soars to over 1.00, CPU usage drops to 0, and the system is unresponsive for 5-30 seconds at a time.

It's always the SSD. Usually the responsible processes are related to my browser (Firefox) but thats because it's my most used application.

That blue line going up indicates "% Highest Active Time" according to the Resource Monitor. This happens even under light load - I could be loading a webpage or opening a speadsheet or what have you.

Here's the pretty sorry-looking specs I get from http://i163.photobucket.com/albums/t299/anon09876/as-ssd-benchINTELSSDSA2MH0811820117-22-54PM.png AS SSD.

The performance degradation of the Intel X25-M G1 SSDs seem to be pretty well documented online, although I don't know if Intel has ever acknowledged it. The most useful article I've come across is this: http://www.pcper.com/article.php?aid=669&type=expert&pid=1 http://www.pcper.com/article.php?aid=669&type=expert&pid=1

I was hoping I maybe just had to update firmware, but Intel's tool says I have the latest (8820) so no hope for a quick fix there. System also tells me I have TRIM on but I don't know if Intel ever updated the G1s to make use of trim.

Other information:

Windows 7 64 bit

8GB RAM

Intel Core 2 Duo CPU (E8500 @ 3.16 Ghz)

And no, I don't try to defrag this drive, I know better than that. It seems to me like the extensive writes to the firefox profile files and maybe my images folder (which I recently moved off the SSD because of all this) wore the drive out - but in only a year! I still have mechanical HDDs from 5 years ago that run indistinguishably from when they were new.

What are my options, besides "wipe with secure erase and reinstall windows", or "buy a G2" - because I'm honestly wary of Intel SSDs now, and maybe SSDs in general. I'm willing to spend money on my PC but dropping $240 ($3/GB!) for less than a year of solid performance from my storage is not worth it. I could have bumped up to a quad core CPU and picked up a fast mechanical HDD for about the price the SSD cost me.

18 REPLIES 18

mmokk
Contributor

p.s. your disk queue lengths are not high at all. in fact they're negligible. look at the y-axis scale on the right, the maximum is 0.5. this means the ssd is easily able to service all requests. the blue line is the queue length, the green is data transfer rate.

this is not to say that you don't have a problem, just to point out that you're not reading the graph correctly.

idata
Esteemed Contributor III

mistermokkori wrote:

p.s. your disk queue lengths are not high at all. in fact they're negligible. look at the y-axis scale on the right, the maximum is 0.5. this means the ssd is easily able to service all requests. the blue line is the queue length, the green is data transfer rate.

this is not to say that you don't have a problem, just to point out that you're not reading the graph correctly.

Well, I have frequently seen the actual Disk Queue Length column under storage surpass 10.00 and Response Time pass 5,000. And when that blue line hits the top (regardless of the number), everything grinds to a halt until it goes back down. So even if the numbers in the examples weren't good illustrations, it's the exact sort of visual I get on the graph every time it dies on me.

This is all generally doing nothing particularly noteworthy. I'm not editing HD video or rendering massive 3D art in photoshop or anything like that. I browse the iternet, listen to mp3s in winamp, chat on IRC, watch streaming video, etc. In short, nothing that should choke a HDD, let alone an expensive SSD.

More importantly, I think, is that I'm doing all the same things I've done for as long as I have owned the drive, and only in the last ~4 months has it caused issues.

jkjkjk wrote:

This is all generally doing nothing particularly noteworthy. I'm not editing HD video or rendering massive 3D art in photoshop or anything like that. I browse the iternet, listen to mp3s in winamp, chat on IRC, watch streaming video, etc. In short, nothing that should choke a HDD, let alone an expensive SSD.

More importantly, I think, is that I'm doing all the same things I've done for as long as I have owned the drive, and only in the last ~4 months has it caused issues.

i agree, the drive should definitely not be acting that way based on your usage pattern. it takes some serious and intentional hammering to get them to the degraded state you're seeing. earlier today i tested a g1 that's about two years old and has had over 3 terabytes of data written to it, it scored around 400 in the as-ssd benchmark.

check the system event log for errors relating to the disk. i would also try swapping out the sata cable. an iffy cable could produce those exact symptoms you've described - long random freezes even during periods of little disk activity. i've seen cheap cables that were merely jostled cause problems. with a conventional hard drive, you'd see these as 'ultradma crc errors' in the smart attributes, but intel ssd's don't keep track of these. but you may still see errors show up in the system event log.

try running the intel ssd toolbox and see what it says about the drive. if you have another pc handy, plug the ssd into it as a secondary and see if it performs any better.

idata
Esteemed Contributor III

jkjkjk, I'm not going to defend Intel's choice to not provide TRIM for G1 drives (I got "stung" for £1,200 for 2 X25E's) but there are things you can do to help yourself out.

If you don't want to reinstall you could try this:

  • Get off that pciide controller and use the Intel controller.
  • Switch to AHCI mode
  • Align the drive (tricky but I think it can be done without a new installation).

By doing those three things you will see a performance jump and they can be done without a new install. (Albeit you will need to change the registry to go to AHCI and install new drivers for the Intel controller).

Next up you could use the PC Perspective trick to reduce degradation. Write a large single file (as close as possible to the free size space) and delete it.

If you decide to image the drive you need to sort it out first. i.e. switch to AHCI, change to the Intel controller and align the drive. (The optimum configuration).

By doing the above you should see an improvement that gets performance back to an acceptable level.

To get optimum performance back and move forward with the right hardware configuration you need to secure erase and do a fresh install. Whilst this may be painful the optimum hardware configuration should go a long way to prevent degradation getting as bad as it currently is.

Try to avoid benchmarking as that will only make things worse.

HDD also requires maintenance and even if you defrag performance will drop over time. I haven't used HDD for some time, but I used to reinstall every 6 months or so to get performance back.