Hardware Reference
In-Depth Information
circumstances exactly? Well, when confronted with questions like that, where you
have
to make a choice one way or another, in general the best thing you can do is to keep it
simple and avoid overthinking the answer. Because real-world testing involves many un-
knownvariables,thesimplestandmostdefinitiveanswerwouldbebasedontherawtrans-
fer speeds alone. In that case, choosing the best answer would be simple: Because USB
2.0 transfers at 480Mbps (60MBps), and 1394a transfers at 400Mbps (50MBps), USB is
faster.
Now I know some people reading this might disagree, including the teacher who gave the
test referenced by the student. Many people believe (and many tests have shown) that,
whereas USBisfaster onpaper,FireWireisfaster inthereal world.AlthoughIhesitate to
makeblanketstatements,IwouldagreethatjustaswithanyinterfaceorbusinaPC,there
are many more factors contributing to performance for a specific task under a specific set
ofconditionsthanjusttherawbusspeedspecification.Althoughmanypeopleassumethat
FireWire is faster in the real world, the truth is that the designs of USB and FireWire are
verydifferent,andonecannotnecessarilypredicttheresultsofaspecificexamplewithout
knowing all the relevant details.
Tohelpmakemypoint,Iransometestsandrecordedtheresults.Ihada7200rpmMaxtor
250MB drive mounted in an external enclosure that supports
both
FireWire 400 (1394a)
and USB 2.0 interfaces. The drive was formatted as a single FAT32 partition and was
about half full. Icreated afolder andcopied a300MB video file into the folder.Ispecific-
ally selected a large file that would not fit in any of the caches or buffers for either the
drive or systems involved.
I used two systems to conduct the tests, and both included FireWire 400 and USB 2.0 in-
terfaces. The systems were as follows:
•
System 1
—Desktop, 3.6GHz Pentium 4 processor, 1GB RAM, Windows XP
•
System 2
—Laptop, 1.7GHz Pentium M processor, 1GB RAM, Windows XP
Totestthethroughput,Icopiedthefileandmeasuredthetimeinsecondsforthecommand
to complete. This meant that the file would be read from and written back to the same
drive,whichamountstoessentiallyatwo-waytripovertheinterface(fromthedrivetothe
PC and back to the drive). I used two commands (
COPY
and
XCOPY
), running each one four
times consecutively, discarding the results of the first run and averaging the other three.
Note
To measure the time it took for each file copy, I used the TimeIt utility, which is a command-
line tool that records the time a specified command takes to run. TimeIt.exe is included with
the Windows Server 2003 Resource Kit Tools, which you can download from Microsoft (
ht-
