Comparing Hard Drive Performance: IDE, SCSI, and RAID

Many people have asked me what hard drive they should
buy; they want to know which ones are the fastest and the most reliable.
Both are good questions, but the answers are not so simple. To make matters
more complicated, Serial ATA drives are coming in the near future and
claim to provide better performance than the IDE Parallel ATA drives that
we all currently use.


There are basically four manufacturers to consider when buying IDE drives:
Maxtor, Western Digital, Seagate, and IBM. Over the years all of these
manufacturers have had their ups and downs, meaning that they’ve all had
their bad production runs — that’s nothing new in the electronics industry
though. Everyone goes through that and as consumers we just have to deal
with it.

Unfortunately IBM’s bad luck has carried on for a bit longer than usual;
I can’t recommend their Deskstar (or as some say, "Deathstar")
hard drives under any circumstances at this time. I’ve seen too many of
them fail miserably over the past year to comfortably say that they are
worth buying. Recently the IBM drive division was sold off to Hitachi,
so perhaps things will improve for them in the future.

Western Digital’s Caviar Special Edition, or what’s sometimes called
the JB series, are good quality drives but I wouldn’t trust any of their
other offerings. Same with Maxtor: anything below 30GB is not worth your
consideration. In fact I would suggest staying away from any 5400RPM drive
by any manufacturer; as a rule they tend to be of a lower quality of manufacture.
As far as warranty is concerned, all IDE drives carry a one-year warranty
except Western Digital whose Caviar Special Edition drives have a three-year
warranty. Despite the shorter warranty, Maxtor drives have a comparable
failure rate to Western Digital.

So now we know where drives stand as far as quality is concerned… but
what about performance? Is one brand faster than the other? What does
that mean in the world of everyday computing? How can we know if SATA
drives are faster if we don’t even know where current ATA IDE drives stand?

Drive performance evaluation

Here are the drives that were available to me for testing:

Brand Model Size Cache Protocol Speed
Western Digital Caviar Special Edition 800JB 80GB 8MB ATA100 7200RPM
Maxtor DiamondMax Plus 9 80GB 8MB ATA133 7200RPM
Maxtor DiamondMax Plus 8 40GB 2MB ATA133 7200RPM
Seagate Barracuda ATA V ST360015A 60GB 2MB ATA100 7200RPM
Seagate Cheetah ST336607LW 37GB 8MB SCSI U320 10000RPM

The system I used for testing was an Iwill P4HT motherboard,
Pentium4 2.4B processor, two sticks of 256MB Apacer PC2700 CL2.5, an MSI
TI4200 video card, an Antec TruePower 330 power supply and Windows XP
Professional. For RAID testing I used an MSI 845PE Max2-FIR motherboard
instead of the Iwill and for the SCSI drive I used an Adaptec 29160 Ultra160
PCI card. Each tested IDE drive was put on its own ATA133 channel on the
Iwill board using an 80-wire IDE cable. The SCSI drive I threw in there
just to show what the real difference between an ATA and a SCSI drive
is.

My testing procedure was as follows: I used a spare 40GB
Maxtor drive to install Windows on and I connected the other drives to
the ATA133 and ATA100 connectors one at a time to test them. For the Maxtor
drives I tested them on both channels to show the difference between the
ATA133 and ATA100 protocols.

For gaming tests I used the built-in benchmarks in Quake3
Arena and Unreal Tournament 2003 using version 1.4 of the HardOCP benchmarking
program. All of the programs were installed to the tested hard drive and
run from that location. Here’s what I found: The game benchmarks showed
such a tiny difference between the best performer and the worst performer
that they aren’t worth publishing — even the RAID arrays were within
a few insignificant frames per second of the other drives. My conclusion
is that no matter what hard drive or drives you use, your gaming performance
will be the same. The only thing that will be altered is the time that
it takes to load each level from the hard drive. For the record: a fast
hard drive will not give you more frames per second in Q3 or UT2003 or
any other 3D game. The exception to that may be if you have a very small
amount of RAM and the game needs to use swapfile space in order to run
— in that case you may see better framereates from a faster drive.

I also used SiSoft
Sandra 2003
to measure the file system’s transfer rate. I ran the
test three times for each drive to make sure that there were no fluke
results, and I recorded the third score only after verifying that it was
consistent with the previous two tests (it always was). It’s important
to note that the same drivers were used for the drive controllers for
each drive. An updated controller driver will likely produce better results
than these, but the purpose of this article is not to show how fast I
can make the drives go; this article’s purpose is to compare drive speed
and efficiency on a common platform. With that in mind, here’s how the
drives performed on my test system from fastest to slowest:

Drive Controller SiSoft Sandra File System Benchmark
2 x 40GB Maxtor DiamondMax Plus 8 ATA133 RAID-0 55966
2 x 80GB Western Digital 800JB ATA133 RAID-0 49350
37GB Seagate Cheetah ST336607LW SCSI Ultra160 46350
80GB Maxtor DiamondMax Plus 9 ATA133 38509
80GB Maxtor DiamondMax Plus 9 ATA100 38024
40GB Maxtor DiamondMax Plus 8 ATA100 35169
80GB Western Digital 800JB ATA133 32303
60GB Seagate Barracuda ST360015A ATA133 27648

We can assume that the Seagate Cheetah would dominate the
charts in RAID-0, followed by the Maxtor 80GB DiamondMax Plus 9 in RAID-0
had I two drives to test with. Likewise a four-drive array would provide
even better performance, but such a controller and four identical hard
drives were also not available to me for testing — besides that, such
equipment is rarely purchased and used for desktop use. The Cheetah drive
is actually an Ultra320, but I didn’t have a U320 controller so it is
only performing at U160 speeds in these tests.

Atto
is a widely recognized hard drive benchmarking utility and it tests the
full range of a hard drive’s data transfer abilities. The scale changes
according to the length of the bar graph, so take care not to confuse
scales when comparing results. The left column indicates the size of the
file that was transferred, which is doubled as the graph descends. I ran
the tests three times to ensure that the results were consistent. Once
again they are rated from fastest to slowest, although I didn’t set up
the WD array again because all of the other results were consistent with
the findings from the Sandra tests:

Maxtor 40GB ATA133 in RAID-0 array

Maxtor 40GB

Seagate Cheetah on SCSI U160 controller

Seagate Cheetah

Maxtor 80GB on ATA133 controller

Maxtor 80GB with UDMA133

Maxtor 80GB on ATA100 controller

Maxtor 80GB with UDMA100

Maxtor 40GB on ATA133 controller

Maxtor 40GB with UDMA133

Western Digital 80GB on ATA133 controller

WD 80GB

Seagate 60GB on ATA133 controller

Seagate 60GB

As you can see there is a measurable difference between the three brands.
But how does that difference show up in real-world scenarios? Every time
your hard drive is running the faster drive will get done faster, obviously.
But your hard drive is rarely stressed for any length of time, and the
differences in the single IDE drives is not significant enough to produce
a noticable difference in performance unless you are working with extremely
large files. In that case you would definitely want to go with a RAID-0
array and you’d want the Maxtor DiamondMax Plus 9 with the 8MB buffer
using an ATA133 controller. If your budget is not quite as limited, any
SCSI U320 drives in an array on a U320 RAID controller will give optimal
performance. Once again I’ll point out that there was no noticable difference
in gaming performance or in everyday computing use between ANY of these
drives — so drive performance doesn’t really matter all that much to
the average user. One thing I did notice is that the 8MB cache didn’t
seem to offer a great advantage, although as you can see above the advantage
is indeed measurable.

Other factors that you may want to consider: the Maxtor drives were noticably
quieter than the Seagate and WD drives, the SCSI drive being the loudest
by far. The Maxtor low-profile 40GB drive ran the hottest out of all of
the drives I tested, although that shouldn’t be a problem for most users
who are on a single hard drive system. If you have more than one hard
drive I would recommend mounting a fan in front of your drive cage to
keep them operating within their specified parameters.

As far as value is concerned, the best value is a toss-up between WD
and Maxtor. The WD offers a longer warranty but it is not as fast. The
Maxtor costs a few dollars more but has a shorter warranty and it has
top performance. The Seagate IDE drives have the least value, being about
as expensive as WD but lacking the performance and the warranty.

Between WD and Maxtor I can’t tell you which one to buy; that’s up to
you. I’ve used both drives in my personal system and I’ve always been
satisfied with their quality and the service I got from the manufacturers.
Don’t let the shorter warranty scare you — Maxtor makes excellent drives.
But if the tiny difference in speed isn’t important to you and you’re
trying to save some money, look for a 40 or 60GB Seagate. In the end you’re
unlikely to regret whatever decision you make.

Discuss this article or get technical support on our forum.

Copyright 2003 Jem Matzan. Verbatim copying and redistribution of this entire article are permitted without royalty in any medium provided this notice is preserved.

About the Author:

Leave A Reply

Your email address will not be published. Required fields are marked *