Real World Tests - Multitasking Performance

To provide a real world example of multitasking we use run Outlook and import 450MB of emails into an account. We then time how long it takes our benchmarking utlity to zip a single 300MB file. To compare our results we calculate the difference between the multitasked process and the single task file zip process.

Outlook + File Zip 1 300MB File
NCQ/TCQ Status
Multitasked
File Zip Only
% Difference
Hitachi T7K250
w/out NCQ
76.543
67.057
12.39%
w/NCQ
79.815
65.641
21.6%
Samsung HD160JJ
w/out NCQ
71.484
58.805
21.6%
w/NCQ
73.554
61.068
20.4%
Western Digital WD1600JS
N/A
74.371
61.182
21.6%

Very disappointing. Hitachi's T7K250 performed better than the rest in this real world multitasking test...with NCQ turned off! Now, drive manufacturers list NCQ as a native feature, but it does not seem to work too well in real world situations. As we look at products in the future, we hope to see atleast some improvement given by Native Command Queuing.

Multitasking Performance - Business Winstone 2004 Thermal and Acoustics
Comments Locked

52 Comments

View All Comments

  • 100proof - Sunday, June 26, 2005 - link

    Above post was for Kristopher.. :D
  • 100proof - Sunday, June 26, 2005 - link

    For all intents and purposes the Sata II designation from my original post was used just for familiarity.. This is how these current drives are advertised and sold, but I'm aware of the name change and did happen to catch the Sata-IO article here on AT. :) Thanks for looking out though :)
  • greekfragma - Sunday, June 26, 2005 - link

    I am still waiting for an answer why they didnt install 6.54 chipset drivers packet instead of 6.39 that is last year's............

    hilarious the ending of the article......i post it once more to be more understood with the things i wrote above.

    "In time, however, we should see improvements in drivers to help reach the true potentials for the new SATA standard."

    plz answer
  • sprockkets - Sunday, June 26, 2005 - link

    Nvm, reading their tech docs state it is 1.5gbits per a second, but figure 80% throughput due to its 8b10b encoding and then divide by 8 for the 150MB xfer rate, so it would assume then the new rate is 300MB from a total of 3.0gb per second.

    It has already been proven NCQ does nothing for the desktop at storagereview.com and decreases performance for non server use. Having twice the bandwidth doesn't help whnen these drives put out a constant 40-70MB anyhow, and burst maybe around over a hundred.
  • sprockkets - Sunday, June 26, 2005 - link

    I don't understand why they say 3.0Gb when the old spec was 150MB a second, shouldn't that mean GB instead of Gb? Doesn't the B being capital mean it should be bytes and not the G?
  • Mariosalice - Sunday, June 26, 2005 - link

    The latest WidowsXP64 Nvidia drivers for AMD are 6.39 and for Windows XP32 are 6.53
  • Mariosalice - Sunday, June 26, 2005 - link

    I would appreciate a SATA II hard disk (raid0 bootable vs single drive vs SATA I)review based on a testbed with dual core Athlon 64, on an nforce4 mobo with windows XP 64 and 6.53 drivers. This is the near future I think.
    As for the tests, just use what you would to compare single vs dual core processors.
    One suggestion. Avast supports 64 bit Windows and you get it for free so you may use it for your tests. Then we may talk how better or not is SATA II compared to SATA I.
  • Olaf van der Spek - Sunday, June 26, 2005 - link

    #34
    > You'd be looking at a minimum of four drives in a striped-array to reach those sort of speeds

    Eh, you do know that SATA is point to point by default and that the interface only becomes a bottleneck if port multipliers are being used, do you?
  • PrinceGaz - Sunday, June 26, 2005 - link

    It should be pretty obvious to all regular AT readers that the increased transfer rate of SATA II will have a negligible effect except with striped or other RAID arrays that can offer over about 150MB/s sustained transfer-rate. You'd be looking at a minimum of four drives in a striped-array to reach those sort of speeds, which is not something most people have. All SATA II does with these drives in any normal situation is allow them to transfer data to and from their 8MB cache a bit quicker, which isn't going to make any noticeable real-world difference.

    I don't have an nVidia chipset so your using the v6.39 chipset drivers didn't mean anything to me, but if they are as old as some here have said, it does throw a big question-mark over whether the results would be the same with the latest drivers.

    As for the benefits of NCQ and the apparent disadvantages of it in many of the tests in this report, all I'll say is Pentium D and Athlon 64 X2. All the disk-intensive multi-tasking benchmarks with those showed major gains with NCQ enabled drives, because the processor was requesting totally seperate data simultaneously while running two threads, whereas a single-core CPU is only running a single task at once and therefore is never truly multi-tasking.

    Good review though, all the graphs worked for me. I'm happy with my Maxtor drives which (touch wood) have served me flawlessly for many years and I've never had a single failure. It's tempting fate saying that but I recently backed everything important up so a drive throwing in the towel wouldn't be a total nightmare.

    I still maintain the major advantage of SATA (I or II) is the smaller cables. I look forward to building my next box where the only ribbon cables have to be routed up to the optical drives.
  • KristopherKubicki - Sunday, June 26, 2005 - link

    100proof: Have you read http://www.anandtech.com/storage/showdoc.aspx?i=24...

    SATAII is a myth....

    Kristopher

Log in

Don't have an account? Sign up now