Version 0.2
Copyright © 2007,2008 by Zack Smith
Introduction
Recently I decided to test the speed of my
various CompactFlash cards and USB 2.0 reader.
Over the years I've collected many CompactFlash cards
but I had never put them all to a systematic test.
So finally, curiosity got the better of me
and I devised a couple tests to run under Linux.
Equipment
I executed my tests on a Toshiba A135 Core Duo based system
running Linux kernel 2.6.15.1.
The CompactFlash cards were put in a PNY
USB 2.0 adapter that was plugged into a USB 2.0
port. There is no model number on the adapter.
Write test
The write test consisted of copying 41.99 MB file
to the USB2-mounted flash drive,
then sync'ing a few times to ensure all data were
written to the drive. I put these commands in a script
and used the "time"
command to time the execution of the script.
Read test
The read test
was very simple. It consisted of running "hdparm -t"
twice and taking a rough average.
Results
Only two of my CF cards actually state on them
what the speed ("X") rating is supposed to be,
so although I've recorded that info, I also calculate
the actual speed.
Note that 1X speed is actually defined as 150 kB/second,
which is 1X for a CDROM drive.
| Brand |
Capacity |
Speed |
Write times (seconds) |
Avg. write time |
Write rate |
Write X speed |
Read rate |
Read X speed |
_ROW
| Lexar |
48M |
4X |
41.418 41.473 41.460 |
average 41.45 |
1.013 MB/s write |
6.753X write |
3.6 MB/s read |
24X |
_ROW
| Sandisk |
64M |
- |
47.863 47.744 47.389 |
average 47.67 |
0.8808 MB/s write |
5.872X write |
1.9 MB/s read |
13X |
_ROW
| PNY |
128M |
- |
29.743 31.023 30.659 |
average 30.48 |
1.378 MB/s write |
9.187X write |
3.8 MB/s read |
25X |
_ROW
| Sandisk |
128M |
- |
23.239 23.358 23.239 |
average 23.28 |
1.804 MB/s write |
12.03X write |
3.8 MB/s read |
25X |
_ROW
| PNY |
256M |
- |
26.552 26.479 26.573 |
average 26.53 |
1.580 MB/s write |
10.53X write |
3.8 MB/s read |
25X |
_ROW
| Sandisk |
256M |
- |
17.326 18.056 17.828 17.671 |
average 17.72 |
2.370 MB/s write |
15.80X write |
3.7 MB/s read |
24X |
_ROW
| Crucial |
256M |
- |
9.755 9.591 9.624 |
average 9.657 |
4.348 MB/s write |
23.97X write |
3.8 MB/s read |
25X |
_ROW
| Lexar |
512M |
4X |
15.116 9.641 10.881 12.626 |
average 12.07 |
3.479 MB/s write |
23.19X write |
4.2 MB/s read |
28X |
_ROW
| Sandisk UltraII |
1024M |
- |
8.786 8.476 8.463 |
average 8.575 |
4.897 MB/s write |
32.65X write |
4.9 MB/s read |
33X |
Discussion
I think the first thing to take note of is that while the PNY
adapter gave good performance, it's not inconcievable that
it has performance limitations. I can't say I'm certain this is
the case until I get my hands on a known-fast CompactFlash card.
The second thing to note is that the 4X rating that appeared
on the two Lexar cards is clearly meaningless. Both cards
performed better. All of the cards performed
better.
Also of note is that fact that I did not do as many
read tests using hdparm as I did write tests.
The reason is that hdparm did not strike me as the best
way to test read speed, but the way I desired to do it
was too tedious. Namely,
my original plan had been to test the read speed
by writing the 42 meg file to the CF card, unmounting it
to make sure the data wasn't cached held in RAM,
then remounting it and time the speed to cat it to /dev/null.
Conclusion
Without knowing what the read/write limitation of the PNY adapter
is, my tests are useful in that they prove
I didn't get ripped off just because I bought unrated CompactFlash
cards: they all performed pretty well.
The next step will be to ascertain the upper limit of the PNY adapter's
performance.
Links
|
|