MaximumPC 2006 01

VERSUSVERSUS

!

VERSUS

!

VERSUS

!

VERSUSV

E

R

S

US

VERSUS

V

E

R

S

US

VERSUS

V

E

R

S

US

!

4 MA XIMUMPC JANUARY 2006

THE TESTS

The Maximum PC testing procedures

for hard drives comprise a delicate

balance of real-world testing and syn-

thetic benchmarks. The fi rst synthetic

benchmark we use is called HDTach 3.0.

HDTach has been around for years and

produces very reliable results. We use it

to measure sequential read speed, which

is the average rate at which data can be

transferred from anywhere on the platter.

Most drives are able to push more than

100MB/s on the outer edges of platters,

but manage just 30MB or so on the inner

rings of the platters, leaving us with an

average in the 50MB range (for 7200rpm

drives, at least). We also use this utility to

measure burst rate and seek time. Burst

rate is also known as “interface speed,”

while seek time is determined by the

drive’s rotational velocity.

Another synthetic benchmark we use

is H2benchw’s Application Index, which

runs a script of seven “real world” applica-

tions, and then offers a score based on the

amount of time it takes the drive to chew

through the script. Our fi nal synthetic test

is IOMeter, which measures how many

requests a drive can fulfi ll in a specifi c

circumstance. For our tests, we load up

a 512-byte read with 32 outstanding

requests, make it 50 percent random and

let her rip. This is as close as you can get

to hard drive torture, while still obeying the

Geneva Convention.

For real-world testing, we measure level

load times in Doom 3 and a 5GB fi le trans-

fer, and we check the drive’s operating tem-

perature sans cooling using Speedfan.

THE RESULTS

As the benchmark chart shows, results

varied wildly on a drive-by-drive basis.

While all four drives sport a now-stan-

dard 16MB buffer, they have vastly

different areal densities due to different

capacities and platter confi gurations.

For example, both the Maxtor

DiamondMax 11 ($350, http://www.maxtor.

com) and Seagate 7200.9 ($355, http://www.

seagate.com) offer 500GB on four platters,

while the Hitachi 7K500 ($355, http://www.hgst.

com) sports a lower areal density with

its fi ve-platter design. The WD400KD

($215, http://www.wdc.com) drive also uses four

Seagate (far left) and Maxtor (far right) have released their 500GB dogs of war to attack Hitachi’s 500GB Deskstar (middle left),

while WD wades into the scrum with its own 400GB Caviar SE. This is going to get ugly!

Best scores are bolded. *The Application Index is a real-world script of six applications. The score is based on the time it takes the drive to complete the scripts. **Hard drive temperatures measured using S.M.A.R.T. data, as reported by the Speedfan utility.

BENCHMARKS

WESTERN DIGITAL WD400KD SEAGATE 7200.9 MAXTOR DIAMAONDMAX 11 HITACHI 7K

HD TACH 3

RANDOM ACCESS TIME (MS) 13.1 13.3 15.5 13.

BURST RATE (MB/S) 138 248 254 133

AVG. SEQUENTIAL READ (MB/S) 57 52.5 59 51.

H2BENCHW

APPLICATION INDEX* 29.7 27.4 30.3 29.

OTHER

DOOM 3 LOADING (SEC) 30 39 33 33

5GB READ (SEC) 101 109 111 102

IOMETER 50 PERCENT RANDOM WORKLOAD (IO/SEC) 230 250 267 244

OPERATING TEMP WITH NO FAN** 52 59 51 63

BEST HIGH-CAPACITY HARD DRIVE

Next-gen, 500 gigabyte drives from Seagate and Maxtor have just arrived in the Lab, so they must defend themselves

against the reigning champs, Western Digital and Hitachi