dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
share rss forum feed


koitsu
Premium,MVM
join:2002-07-16
Mountain View, CA
kudos:23
reply to dolphins

Re: [hard drive] CDM tests have me worried...

Information in question isn't that helpful, sorry to say. Benchmarks don't give necessarily any indication as to what the system is doing -- safe mode helps but only to some degree (I believe a lot of Services are still started in safe mode).

It sounds to me like something else in Windows was doing a large amount of disk I/O (and not necessarily Photoshop).

It looks like you're using Windows XP, which makes tracking down what programs are doing massive I/O a little more difficult. Sadly, this is one of the things Windows 7's Resource Manager is incredibly good at. The only thing on XP that I can think of (hope others have better recommendations) that can do this is perfmon/Performance Monitor, but I don't think that will show what process is doing the I/O, just the rate of I/O.

Regarding the benchmarks: I see no anomalies. You can't seriously be worried about a ~7MByte/sec delta on read and write speed amongst 4 tests, can you? Min/max read=125MB/132MB, min/max write=83MB/89MB. That's quite acceptable.

SMART attributes would be a better indicator (for me anyway) to determine if there is something truly amiss on a hardware level.

Finally, it's very important to remember that MHDDs tend to gradually decrease in performance over time. As I understand it, this has to do with the magnetic aspect of drives; as the drive ages over time, thus more and more writes are issued over time, the substrate begins (to some degree) have an increased number of errors. These aren't errors you would see in the OS (as in I/O errors), these are purely substrate-level and are hidden from the user (drive mechanics, firmware, etc. deal with all of this transparently). I can attest to this theory being fairly true; I've had WD Black drives which performed fantastic when I bought them (start of drive: read=165MB/sec, write=155MBytes/sec), but after 2 years of constant use began performing more along the lines of read=130MB/sec, write=110MByte/sec.. I have some drives (different brand) which are nearly 6 years old and the performance on those is abysmal compared to when I got them.

If your SMART attributes look OK (to me, after I see them), then I'm going to recommend this thread be moved to the Microsoft forum, as folks there can definitely help you with figuring out what's doing disk I/O. One thing I can think of, for example, is a highly fragmented page file (which is something most filesystem defrag utilities do not defrag; there are other utilities, including some from Microsoft/Sysinternals, that can do this for you). The kernel takes care of managing this on its own, and if at some point in the past memory pressure was quite high, I can see this being a potential (but not guaranteed) problem.

{babbling koitsu}
I guess you could say this is another reason why SSDs are nice -- not that NAND flash lasts forever (hardly so, but wear levelling works wonderfully as long as you treat your drive well), but just that it's not going to "perform worse" over time as long as the manufacturer didn't write their firmware like complete crap. With SSDs TRIM is pretty important (GC is generally okay but TRIM ensures good performance all the time), and since I myself use XP with an SSD (thus no TRIM support), I naturally run the Intel SSD utility every ~2 weeks to make sure everything is in good shape. My Intel 510 120GB still performs just as good as it did when I got it a few years ago -- I know because I did benchmarks a few months ago before and after a Secure Erase and the performance was the same.
{/babbling}
--
Making life hard for others since 1977.
I speak for myself and not my employer/affiliates of my employer.