Test Setup - Hardware

We have changed our test beds to reflect changes in the current marketplace. While we wanted to change to the AMD AM2 platform, the continual product delays forced us into staying with a socket 939 based system. Based upon the continuing proliferation of dual core processors along with future roadmaps from AMD and Intel signifying the end of the single core processor on the desktop in the near future, we have upgraded from our AMD Athlon64 3500+ to an AMD Opteron 170. This change will also allow us to expand our real world multitasking benchmarks in the near future. We will review our test bed requirements once we have an opportunity to thoroughly test the AM2 and Intel Core 2 Duo platforms.

We debated on the memory size for our IPEAK trace file creations and decided to move to 2GB of system memory. A system with a 1GB memory configuration is the predominant standard at this time in the enthusiast community, but 2GB memory setups are fast becoming the future standard. Although a 1GB memory installation allows us to capture and report a higher amount of disk activity in certain applications; we decided to make the switch at this time as the performance difference is minimal when compared to the 1GB trace files.

Standard Test Bed
Playback of iPEAK Trace Files and Test Application Results
Processor: AMD Opteron 170 utilized for all tests.
RAM: 2 x 1GB Corsair 3500LL PRO
Settings: DDR400 at (2.5-3-3-7*c* 1T)
OS Hard Drive: 1 x Maxtor MaXLine III 7L300S0 300GB 7200 RPM SATA (16MB Buffer)
System Platform Drivers: NVIDIA Platform Driver - 6.85
Video Card: 1 x Asus 7600GS (PCI Express) for all tests.
Video Drivers: NVIDIA nForce 84.21 WHQL
Optical Drive: BenQ DW1640
Cooling: Zalman CNPS9500
Power Supply: OCZ GamexStream 700W
Case: Gigabyte 3D Aurora
Operating System(s): Windows XP Professional SP2
Motherboards: MSI K8N Diamond Plus


Standard Test Bed
Creation of iPEAK Trace Files
Processor: AMD Opteron 170 utilized for all tests.
RAM: 2 x 1GB Corsair 3500LL PRO
Settings: DDR400 at (2.5-3-3-7*c* 1T)
OS Hard Drive: 1 x Maxtor MaXLine III 7L300S0 300GB 7200 RPM SATA (16MB Buffer)
System Platform Drivers: ATI Platform Driver - 1.1.0.0
Video Card: 1 x Asus 7600GS (PCI Express) for all tests.
Video Drivers: NVIDIA nForce 84.21 WHQL
Optical Drive: BenQ DW1640
Cooling: Zalman CNPS9500
Power Supply: OCZ GamexStream 700W
Case: Gigabyte 3D Aurora
Operating System(s): Windows XP Professional SP2
Motherboards: ECS KA1 MVP Extreme


We chose the ECS KA1-MVP as the platform for creating our IPEAK trace files. This affords us an updated system with the capability of correctly creating and storing our trace files on a SATA based drive. It also allows us to utilize a modern video card for the game play trace results which are captured with the graphic settings at a typical 1280x1024 resolution.

You may have noticed we did not use the MSI K8N Diamond Plus for both purposes, though the balance of the component choices are essentially the same. We experienced inconsistencies with our trace files on this platform, our ULi M1575 or M1697 boards, and those of any Intel based systems featuring the ICH6 or ICH7 chipsets. The ATI SB450 proved to be the only currently available chipset that produced repeatable results on all platforms when utilizing the IPEAK WinTrace32 program. Note that this is a common issue with IPEAK: once you create trace files that perform consistently, they will work fine on any platform, but creating the trace files requires the use of specific platforms/drives with prior trace files being developed on an Intel board with the ICH5 Southbridge.

Test Setup - Software Features and HD Tach Test
Comments Locked

44 Comments

View All Comments

  • Larso - Monday, May 22, 2006 - link

    quote:

    The pictures for the noise level are wrong. You put the dbA level as if it was a linear scale. It's not that way, the space between 0dB and 10dB should be smaller than the space between 10dB and 20dB. That way it will show more clearly the difference between the noise levels. It's a logarithmic scale.

    Well, you usually plot dB on a linear axis, taken for granted that the interpreter knows the exponential nature of decibels (+10 dB sounds like doubling the loudness). But since our ears/brain also interpret sound levels in an exponential fashion, its not _that_ misleading...

    If you really want to show the soundlevel in a linear fashion, you would have to measure loudness in http://en.wikipedia.org/wiki/Sone">Sone

    - Gary: Sorry if I've missed it, when you measured the loudness, was the drive secured to the case with screws, or was it suspended softly? It would be interesting to know how much of the noise was due to vibration, and how much was actual noise emission from the drive.

    Perhaps some kind of soft suspension would make the drive noise more bearable, subjectively?
  • Gary Key - Thursday, May 25, 2006 - link

    quote:

    - Gary: Sorry if I've missed it, when you measured the loudness, was the drive secured to the case with screws, or was it suspended softly? It would be interesting to know how much of the noise was due to vibration, and how much was actual noise emission from the drive.


    The drives are suspended via soft rubber bushings in the drive cage in order to equalize the test results between drives (as much as possible). This drive will create an additional low pitch vibration when attached directly but nothing like the sound a Raptor produces when attached directly, subjective opinion of course. :)
  • TonyB - Thursday, May 18, 2006 - link

    how soon till we see 1TB hard drive. 7200.11?
  • Gary Key - Thursday, May 18, 2006 - link

    We should see a 960GB drive by Q4. We will see the 1TB+ drives in 2007, maybe earlier depending on how well the new 200GB plus platters test out. Also, if Hitachi or WD press Seagate on the issue in Q4 I am sure it be out early. ;->
  • Zaitsev - Thursday, May 18, 2006 - link

    Page 4 reads, "we decided the make the switch at this time as the performance."

    I enjoyed the article, Gary.

    "I would much rather show the benefits of RAID 5, 0+1, 10 in a separate article"

    I think that's a great idea, and am looking forward to it.

  • Gary Key - Thursday, May 18, 2006 - link

    quote:

    Page 4 reads, "we decided the make the switch at this time as the performance."


    The link highlight on the word "applications" right before "we" is hiding the semi-colon somewhat. I will see if we can edit the hyperlink to correct this. :)

    Thanks for the comments. We are trying to head in a more technical direction in the storage section. We have some additional audio/video tests coming along with application timer benchmarks that should lighten up the presentation. Also, we did not report WB99, IPEAK Access, Disk Bench, IOMeter, or other benchmarks yet, still gauging what information is valuable, interesting, and required.
  • JarredWalton - Thursday, May 18, 2006 - link

    Actually, he was referring to the word "the" - fixed now, as it should have been "to". :)
  • Zaitsev - Thursday, May 18, 2006 - link

    Sorry about that. I will be more clear next time. :)
  • Gary Key - Thursday, May 18, 2006 - link

    quote:

    Sorry about that. I will be more clear next time. :)


    I will wear my glasses next time.... LOL
  • peternelson - Thursday, May 18, 2006 - link


    Hi,

    I can understand why you want to test sound levels as close as 5mm.

    However, to someone quickly looking at the results charts without carefully reading the text they might think your db(A)@5mm are comparable with db(A) measurements taken at more conventional distances like 1 metre away.

    To avoid such errors being made when comparing BETWEEN reviews, please clearly label the audio charts not just "db" but "db(A)@5mm"

Log in

Don't have an account? Sign up now