The Test

To benchmark on our selected CompactFlash media, we picked up a Sabrent SBT-CRW4242-in-1 flash media reader that has an external USB 2.0 interface. Take a look at the test bed below.

Our test bed:

AMD Athlon 64 3500+ (2.2Ghz)
Giga-byte GA-K8NXP-SLI
Western Digital WD1600JS
NVIDIA 6600GT SLI Edition (single 128MB card )
1GB (512MBx2) Corsair XMS4400

Sabrent SBT-CRW42 42-in-1 reader (Integrated Circuit Solutions, Inc. based hardware)

Our motherboard is an nForce4 based board, which features support for the SATA II standard, up to 3.0 Gb/sec SATA transfer rates, NCQ and TCQ , and the USB 2.0 standard for up to 480Mb/sec transfer rates.

We used the following NVIDIA platform drivers in conjunction with our test bed:

nForce4 Chipset Driver 6.82
NVIDIA graphics driver 81.95
Windows XP SP2 w/out further updates

 AnandTech Flash Media Tests
Real World File System Task Tests timed tests of copying files to and from flash media
HDTach 3 RW Synthetic test to measure average read and write rates of flash media
SiSoft Sandra Removable Storage/Flash Media Benchmark

*Each card is formatted for the FAT32 file system using the default cluster size, since the majority of devices only support this file system.

Index Real World File System Tasks
Comments Locked

24 Comments

View All Comments

  • Anton74 - Saturday, December 24, 2005 - link

    quote:

    expected life expectancy

    It should be noted, of course, that I am a representative of the Department of Redundancy Department. ;-)
  • heulenwolf - Friday, December 23, 2005 - link

    At my workplace, we use CF cards for all sorts of things that don't involve cameras. I, for one, am happy to see random access tests. I agree that they should be in addition to sequential tests to give camera users some info.

    I agree that the graph types need some work. Specifically, the spline (or whatever interpolation function it is) used on page three, "Real World File Copy Test to Media," connects dots between different cards. This interpolation tells us nothing and only confuses the situation by connecting what should be disassociated information. For example, there's no information provided by looking at what the spline says occurs "half way" between an Edge and a Kingston card. If Purav were to switch the graphing inputs around such that the X-axis is file size and make separate lines for each card, then, assuming he'd picked an appropriate interpolation function for the "Real World" transfer times, we might be able to tell something about transfer times for file sizes in between the three tested. If not, then drop the spline and just plot the dots.
  • Lifted - Friday, December 23, 2005 - link

    So where are the RAID benchmarks?
  • mindless1 - Friday, December 23, 2005 - link

    "Special thanks to NewEgg for providing us with the CompactFlash cards for this review."

    Yes, thank you Newegg. I love seeing reviews where Newegg donated parts, since so many of us shop there it seems a win-win for everyone.

Log in

Don't have an account? Sign up now