SATA-roundup – 160 GB

We’ll begin this harddrive review just as we did the last one, we’ve been inactive in covering the storage section for some time now, but now comes the time to change that….

We’ll begin this harddrive review just as we did the last one, we’ve

been inactive in covering the storage section for some time now, but now comes

the time to change that. As a preview of what you’re about to see today I can

tell you we’re about to have a closer look at a couple of mainstream harddrives

using the SATA interface. We’re talking about drives with 160 GB capacity, 7200

rpm speed and 8 MB cache. Harddrives with 120 GB storage capacity have long

been the standard and most bang for the buck (most GB/penny) but disks with

160 GB capacity have taken that lead and today we thought we’d do a straight

forward shoot-out in this price range.

The list of manufacturers represented in the test is almost complete, Hitachi, Samsung, Seagate and Western Digital. Regretfully we were unable to acquire a corresponding disk from Maxtor. We’ve tried, for a long time, to get a hold of one from several stores, distributors and even themselves but to no avail, it’s out of stock everywhere. On the other hand we managed to get our hands on one of Western Digitals latest performance monsters, the Raptor 74 GB. We’ll be testing this one as well together with the other disks as a reference.

Before we continue with the review we thought we’d cover some basic stuff on the now not so new SATA interface, just click the link to continue.

---

As an evolutionary successor to PATA (Parallel ATA), a new interface for storage devices appeared about a year and a half ago, SATA (Serial ATA). In August 2001, a large syndicate of companies decided to develop this new interface to replace the old one that has been around for twenty years, an amazing amount of time for any type of technology to still exist in such a rapidly developing industry. The aim with the new interface was to combine the benefits of PATA and SCSI (Small Computer System Interface).

Below you will find

a table with the properties that the different interfaces have. More on the

properties to follow. Note that it’s SATA 1.0 we are talking about here unless

otherwise stated.

Feature	Parallel ATA	Serial ATA	Parallel SCSI
Low Price	X	X	–
Point-point connection	–	X	–
Cyclic Redundancy Check (CRC) commands	–	X	X
Hot-plug/Hot-swap support	–	X	X
Efficient cables, connections, and back plane	–	X	–

One thing to mention,

however, is that SCSI drives have several benefits that SATA doesn’t. They have,

for example, upwards 15k rpm spindles, controllers that support up to 320 MB/s

transfers, access times close to half that of a PATA/SATA, and they often come

with a 5-year warranty. But all this has a price. SCSI solutions are much more

costly and are therefore not suited for the average user.

Point-point connection: This is an important feature

that increases the performance and dependability compared to the split connections

of the PATA and (also parallel) SCSI. Point-point connection means exactly what

it says; every port on the SATA controller is connected to one unit only. See

the illustration below.

Since the bus isn’t shared, each unit can always communicate directly

with the system. This means that all the available bandwidth is dedicated to

every unit at the same time. This way of connecting eliminates what is commonly

known as "arbitral delay " which is some times associated

with bus sharing. This type of connection also alleviates the installation of

disks since one can ignore the master/slave or disk-ID (SCSI) considerations.

Cyclic Redundancy Check (CRC) at commands: Probably

the greatest improvement over PATA is the cyclic redundancy check (CRC) at commands.

Both SATA and PATA have CRC, but it differs slightly with the latter. Whereas

PATA only implements CRC for the data transfer, SATA also adds it to the commands.

The CRC is a kind of supervision that ensures the data, and now also the commands,

are done right. This leads to improved transmissions, and the disk simply becomes

more reliable by the disk always making sure everything is received correctly.

Hot-plug/Hot-swap support: This has been a standard

with he SCSI interface for a long time which enables removing and replacing

disks without having to shut the system down. This is important with servers

and other larger systems that cannot be shut down. Nothing that your average

Joe really needs, but with the implementation of this function, SATA disks become

a good alternative to SCSI for the business sector.

Efficient cables, connections and back plane: The

first thing one notices about the SATA cable is its size, it’s so small and

petite! For the end-user this will probably feel like one of the biggest advantages

with SATA, namely the absence of the clumsy IDE cables. The thin SATA cables

give a much better air flow in the case, a much desired effect considering the

amount of heat produced by today’s hardware. Improved air circulation leads

to a more stable system. Another thing worth mentioning is that the SATA cables

can be up to 1 meter long, while the IDE cable is limited to 0.9 meters.

SATA cable from above, from the side, round IDE cable, and regular IDE cable.

The L-shaped data- and power connectors of SATA make it impossible

for the user to connect them the wrong way.

SATA data connector	SATA power connector

Besides what has been mentioned above, we should add that SATA has

a much lower signal voltage (0.5 V max) compared to PATA (5 V max). This leads

to a lower power consumption and less electromagnetic interference. For a private

user with maybe one or two disks, the savings made by switching to SATA will

most likely not be noticeable, but for larger companies and others with a need

for several disks, the scenario might be different

The Future – Serial ATA 2.0

The development of this "new" interface continues, and

not too long ago, the Serial ATA syndicate announced some specifications and

other information regarding the even newer interface. Serial ATA 2.0 is already

under development. What does it have to offer?

There are mainly two points that are new:

1. The transmission speed has been doubled to 300 MB/s

2. New cables and connections to enable other implementations and models of

usage.

SATA 2.0 will not require any new cables, instead, new cables and connectors

will appear for other solutions , e.g. external. On the other hand, there will

be a second generation SATA 2.0 (alternately called SATA 3.0) that will appear

some time in 2007. This second generation will have, are you sitting down?,

yet another doubled transmission speed, 600 MB/s(!). Cables and connectors may

be upgraded with the second generation, but that remains to be seen. I would

also believe that other features will be introduced with the second generation

of SATA 2.0, but in order to use the extreme transmission speed, something extraordinary

will have to happen on the mechanical side of the hard drives. Since this is

something so far into the future, we’ll have to take it with a pinch of salt,

but it definitely looks very exciting.

We’re looking forward to the end of this year/beginning of next when the new

interface should have hit the market. It seems as though some hard drives will

show up with a partial list of SATA 2.0 specifications as early as the third

Quarter of this year. We’ll naturally keep you posted!

If you want more information, head on over to the Serial-ATA syndicates

home page. Let’s continue with the review.

---

Hitachi Deskstar 7K250

Hitachi have been stalked by the reputation gained from the IBM GXP series of disks for a long time, some of which might not have been the best representatives of the company. We’ve however gotten a hold of an example from their far better series the 7K250, the specifications are available below.

Specifications: Hitachi Deskstar 7K250
Model number	HDS722516VLSA80
Firmware	V340A60A
Storage capacity	164.7GB
Rotational speed	7 200RPM
Cache size	8MB
Form factor	3.5″
Seek time	8.5ms
Average latency	4.17ms
Number of platters	2
Number of reading heads	4
Maximum external transfer speed	150 Mbyte/s
Maximum internal transfer speed	757 Mbit/s
Noise level	28-? dB (only idle specified)
Interface	SATA 1.0
Warranty	3 years
Price	~1050 SEK (Exchange rates)

If we have a look at the specifications we can see this drive has a slight increase in storage capacity compared to the competitors, 164,7 GB instead of 160GB. That’s not much, but 4.7GB is always 4.7 GB.

Top view

Bottom view

The upper side can’t be called anything but dull, but that’s far from

unusual in this area of hardware. The lower side, the PCB to be more exact,

is something we’re going to have a closer look at though. We can clearly see an

IBM chip in the middle of it. That means Hitachi still hasn’t broken completely

free from IBM yet, this shouldn’t be something to be alarmed about though.

Hitachi choose to use Infineon memory for the hard drive cache.

The chip with a M on it which you can see in the lower right is a chip from Marvel, a serial ATA bridge to be particular. This means this isn’t a “real” SATA drive if you want to put it that way. It receives serial signals but converts these to parallel ones. We’d naturally be more happy to see a native SATA chip, but as it turns out today you actually don’t lose any performance with one of these bridges. A positive side effect with this chip is that it handles ATA Command Queuing (CQ). This might sound just fine, but in order to utilize this your controller will have to support this feature as well. As of today there are no commercially available SATA controllers that support this function (there are PCI-X controllers with this feature, but it’ll probably take a while before we see them widely spread on the market). This all boils down to another sad fact, as this is actually a PATA drive, it still won’t work unless the controller sends emulated PATA CQ signals instead of ordinary CQ commands. We can thereby conclude that as it stands today there is basically no benefit what-so-ever of having this feature, but it does make the drive a tad bit more future compatible.

As fast as we get the opportunity we’ll do another review where we cover the possible benefits CQ might bring to hard drive performance. That will have to wait a while due to the reasons stated above however.

Back panel

Another proof that this drive isn’t a “real” SATA drive, the 12V connector. As you read in the last page it’s the L-shaped power connector that belongs to serial ATA. As previously mentioned this doesn’t however affect performance as of today and therefore it feels sort of comfortable to have it as far from everybody packs a power supply with the necessary SATA power cords. As you can see in the picture Hitachi attached a sticker with a warning label on it. It states you should only connect one power cable and not one to each of them as this might result in unwanted effects. This should be present on all drives with dual power connectors but it’s always nice to see an effort taken in order to make the customer aware of this in order to minimize the risk of damage to the drive.

---

Samsung SpinPoint P80

Next one up is Samsung.

Specifications: Samsung SpinPoint P80
Model number	SP1614C
Firmware	SW100-27
Storage capacity	160GB
Rotational speed	7 200RPM
Cache size	8MB
Form factor	3.5″
Access time	8.9ms
Average latency	4.17ms
Number of platters	2
Number of reading heads	4
Maximum external transfer speed	150 Mbyte/s
Maximum internal transfer speed	840 Mbit/s
Noise level	27-? dB (Only idle specified)
Interface	SATA 1.0
Warranty	3 years
Price	~1100 SEK (Exchange rates)

The P80 has the

highest maximum internal transfer speed among todays 160GB discs. More than

that the specifications doesn’t tell.

Top view

Bottom view

The top of the Samsung does actually look any different than all the other

drives today, just a little bit more round than the others. We think it’s good

to se a little variation even if this is not very important. It’s a very

smooth and easy way of holding the drive while installing it.

On the circuit board we find two chips from Marvel, the big one in the middle

(88i6522) and the smaller one a little bit further down. The big is the hard

drive controller who actually only supports ATA-133 and the smaller one is the

same chip as on the Hitachi disc, but a later revision (B). This is in other

words also a PATA disk. The harddrives cache is what you se down to the right

and it is manufactured by Samsung.

Back panel

Unfortunately, Samsung has chosen not to put a regular 12V-contact on their

back panel. The pins you see without any jumpers you can ignore, they have no

function.

One thing to add is that Samsung sends a small installation guide and a couple

of screws with their disks, something very positive we think. Let’s go

on to Seagate’s contribution.

---

Seagate Barracuda 7200.7

Next out is Seagate, with the following specifications below.

Specifications: Seagate Barracuda 7200.7
Model number	ST3160023AS
Firmware	3.18
Storage capacity	160GB
Rotational speed	7 200RPM
Cache size	8MB
Form factor	3.5″
Seek time	8.5ms
Average latency	4.16ms
Number of platters	2
Number of reading heads	4
Maximum external transfer speed	150 Mbyte/s
Maximum internal transfer speed	683 Mbit/s
Noise level	27-34 dB
Interface	SATA 1.0
Warranty	3 years
Price	~1150 SEK (Exchange rates)

The specifications

for Seagate’s harddrive are nothing out of the ordinary. Its access time is

among the lowest of todays 160GB-HDD’s, together with Hitachi’s. The main reason

people choose Seagate is due to their good reputation when it comes to noise

level. Even though the noise specification isn’t that special; 25dB is the lowest in the review but

34dB at full load is hardly a reason to start celebrating.

Top-view

Bottom-view

It’s time to take a look at the lower side. Seagate uses memory

from Hynix for their cache, which can be examined on the upper right picture.

Beside this there’s a chip from LSI Logic on the PCB that catches the attention.

This is not the regular SATA-bridge which you might believe if you’ve read the

recent pages, but a native SATA controller. This means that the hard drive

from Seagate is the only "real" Serial ATA hard drive in todays reviews,

and also the only company which manufactures real Serial ATA hard drives in this

segment.

Back panel

There is nothing much to say about its back panel, one power connector

and a SATA-connector.

---

Western Digital Caviar SE

The next drive in todays tests come from Western Digital. According to customs, here’s the specifications.

Specifications: Western Digital Caviar SE
Model number	WD1600JD
Firmware	02.05D02
Storage capacity	160GB
Rotational speed	7 200RPM
Cache size	8MB
Form factor	3.5″
Seek time	8.9ms
Average latency	4.2ms
Number of platters	2
Number of reading heads	4
Maximum external transfer speed	150 Mbyte/s
Maximum internal transfer speed	748 Mbit/s
Noise level	33-35 dB
Interface	SATA 1.0
Warranty	3 years
Price	~1150 SEK (Exchange rates)

There’s not much to add to the specifications either. A rather high noise level specified though, especially when idle.

Top view

Bottom view

There’s not much to say here either as Caviar SE’s upper side is

quite modest and the PCB turned inward. Due to that we can’t see what chip and

memory this drive is using. On the other hand this is a pretty smart move by

WD since it’s somewhat safer to keep the chips on the inside rather than on

the outside. I do however suspect that this PCB is equipped with the Marvel

chip found on the Hitachi and Samsung (and also on the WD Raptor 74 GB) drives

since the unit has double power connectors (no native SATA). WD also informs

the user not to connect both the power connectors at the same time on the upper

side of the drive.

Back panel

The back panel looks almost like any other in the review, one 12V connector, one serial data connector, one SATA power connector and a couple of unused pins. Almost in this case is due to the two small holes with a kind of clip visible next to the SATA connectors. These are holes for a new cable that WD developed. It’s called SecureConnect and combines both power and SATA connector into one. The L-shaped connectors have a small drawback in that they’re not too firm in their connections, WD has however remedied this with this new cable. Alas, we didn’t receive such a cable for todays review.

Western Digital Raptor 74GB >

---

Western Digital Raptor 74GB

The last and the least (when it comes to storage capacity), Western Digital’s

latest performance drive, the Raptor 74 GB. We said that this would be a 160 GB

SATA roundup, but we took the opportunity to include this performance monster

from Western Digital, which is also used as a reference drive.

Specifications: Western Digital Raptor 74GB
Model number	WD740GD
Firmware	21.08V21
Storage capacity	74GB
Rotational speed	10 000RPM
Cache size	8MB
Size	3.5″
Search time	4.5ms
Average latency	2.99ms
Number of platters	2 st
Number of reading heads	4 st
Maximum external transfer speed	150 Mbyte/s
Maximum internal transfer speed	102 Mbyte/s
Noise level	32-36 dB
Interface	SATA 1.0
Warranty	5 years
Price	~2200 SEK (exchange rates)

High rotational speed, low search time, low latency, high internal transfer rate and a 5 year warranty. It’s the Raptor coming.

Top

Bottom

Another drive with a different design. The reason for Western Digital to make

one side almost a heatsink with fins is probably because it results in better heat

emission. The circuit board has been turned over except for one chip, known

as The Marvel chip, here seen in a later version (C).

Rear panel

Identical to the previous drive.

Test system and controllers >

---

Test system and controllers

Test system
Hardware
Processor:	Pentium 4 2.4GHz (533MHz FSB)
Mainboard:	Abit IC7-MAX3
RAM:	2 x 256MB Apacer PC2700
Graphics card:	Hercules Radeon 9800Pro 128MB
Harddrive:	Seagate Barracuda 7200.7, 160GB, ST3160023AS Western Digital Caviar SE, 160GB, WD1600JD Hitachi Deskstar 7K250, 160GB, HDS722516VLSA80 Samsung SpinPoint P, 160GB, SP1614C Western Digital Raptor, 74GB, WD740GD Western Digital Raptor, 36GB, WD360GD
Controllers:	SATA: Integrated controller in the south bridge ICH5R PCI Chip on the motherboard by Silicon Image (SiI3114) PCI card Promise Technology SATA150 TX4
Sound card:	Creative Soundblaster Live! 5.1 Platinum
PSU:	FSP 350W
Ethernet:	Intel PRO/1000 CT
Software
Operating system:	Windows XP Professional (Service Pack 1 + updates)
Resolution:	1152x864x32bit, 100Hz
Drivers:	Catalyst 4.5 DirectX 9.0b
Benchmarking Software:	FC-Test v0.5.3 HDTach 2.61 HDTach 2.70 PCMark04 v1.2.0 Winbench99 v2.0

Before we go on

with the review we’re going to say a few words about the different SATA controllers.

Intel ICH5R

An interesting

characteristic with the Intel 875P chipset is its south bridge with integrated

SATA controller. The controller is entirely independent and doesn’t rely on

the PCI bus, which other integrated controller chipsets on mainboards do. The

advantage of having the controller integrated in the south bridge is that no

data has to go via the PCI bus. This should lead to better transfer speeds,

most of all in RAID where the PCI bus really is a bottle neck. But today it’s

not RAID that we’re testing so we’ll see what the south bridge can do. The south

bridge offers two ports and they support RAID 0/1, but unfortunately not RAID

0+1.

Silicon Image SiI3114

Silicon Image seems

to be really popular among mainboard manufacturers. Their chip is integrated

on massive amounts of motherboards. As you can see, this controller is the type

that was mentioned above, an integrated PCI chip. This chip offers four ports

and they support RAID 0/1/0+1.

Promise Technology SATA150 TX4

Since many of you

at home maybe don’t have mainboards with integrated controller chips of any

kind, we’ve also chosen to use a PCI controller card. Promise Technology manufactures

very competent and cards worth its price of this type and they were kind enough

to send us this card which is a pure SATA controller card with four ports.

In other words

we’ll be testing this review’s drives with three different controllers. This

is for making a more fair judgment, depending on what products you readers

have bought or looking to buy, but also look at it as a little test for the

controllers themselves.

There’s only one

more thing we’d like to say before we go on with the tests. We used Hitachi

Feature Tool v1.94 to set AAM (Automatic Acoustic Management) to maximum performance

(Normal Seek Mode) on all drives (except for Seagate. Their drives don’t support

it because of some lost court case). You can read more about the tool here.

Now we’ll continue.

---

Test system and benchmarks

The first program

we’ve included in the benchmarks is FutureMark’s well-known PCMark04. PCMark02

was completely rejected since it had a tendency to repeat results and sometimes

even give completely erroneous ones. With the introduction of PCMark04 the picture

is completely different, it stands out to be a good tool to measure performance.

We are going to use 3 out of 4 available HDD-test offered by this program. These

tests are based upon four common operations all users perform. PCMark04 simulates,

in other words does the same operations, as the fields of these tests represent

and measures the performance. Information about the various test can be found

below.

Windows XP Startup:

This simulation is based upon the operations needed to startup Windows XP.

Application Loading:

As the name indicates, it measures the loadtime of different programs. The set

includes opening and closing of the following programs; Microsoft Word, Microsoft

Word, Adobe Acrobat Reader 5, Windows Media Player, 3DMark 2001SE, Leadtek Winfast

DVD and Mozilla Internet Browser.

File Copying: Basic

file copying of a file approximately 400 MB in size (not specifically specified).

General Hard Disk

Drive Usage: Operation that includes harddrive operations during a majority

of common programs. These are:

– Opening of a Microsoft Word-document, accomplish grammars, saving the

document and closing the program.

– Zipping and Unzipping using WinZip.

– File encryption and decryption using PowerCrypt.

– Virus scan of files with F-Secure Antivirus

– Mp3-playback using Winamp

– WAV-playback using Winamp

– DivX-video playback using a DivX codec and Windows Media Player

– WMV-video playback using Windows Media Player

– Picture viewing in Windows Picture Viewer

– Internet browsing through Microsoft Internet Explorer

– Gametesting through loading, playing and terminating the game Ubisoft

Tom Clancy’s Ghost Recon

As I said a bit earlier

we are going to use three out of the four available tests. Why not all four?

Well, the copy-benchmarking in PCMark04 doesn’t seem at all reliable and due

to this we are excluding this test. Partially because it’s unsound to base a

copy-test all upon one big file, and partially since the test fundaments on

copying a file from the drive you’ve installed the program to the partition

you’re testing. This means that not only is the harddrive you’re testing the

dependable factor, but also your systemdisk or where ever you have installed

PCMark04. Due to this we’ve chosen to use a different program this matter, but

more about that on the next page.

In this test we used a 32 GB parition at the very beginning of the harddrive

(recommended by FutureMark) which we formatted using NTFS. Our benchmarks were

performed 3 times on each harddrive, and then calculating the average value

which are shown below.

We start out with

the Windows XP Startup test.

Nothing odd here,

the only thing that astonishes is that Hitachin manages to pass Raptor 36GB

in the first two benchmarks. All harddrives performs best with the Promise controller

card in this test. The reason why Samsung and WD are the last of them all is

explained by their slow access times, which is an important aspect in this benchmark.

It also explains why Hitachi scores well.

The trend continues. Hitachin performs best, though the upper hand

have decreased this time. Again each and all disks achieves the best performance

with the Promise controller card.

The harddrive-, and controller patterns we’ve seen earlier seems

to repeat itself. As a rundown of this benchmark we can say that the Hitachi

prospered among today’s mainstream-disks when it comes to performance. In these

test we also see the true strength of the Raptors, especially through the Promise

controller. WD’s Raptor 36GB seems to have some kind of problem, especially

with ICH5R, we’ll have to see if this occurs again. One thing that surprised

us was that i875P’s integrated SATA-controller in the southbridge seems to perform

worse than Promise’s controller card in all PCMark04-tests. ICH5R and Sil3114

performs quite equally.

---

Winbench99 Version 2.0

Winbench99 version

2.0 is a benchmarking utility that is getting a bit old, but still works very good.

The functions are similar to those in PCMark04; you “record” how a

number of applications use the harddrive and this is reflected in the tests.

I suppose almost all of you have used this software at some occasion and will

not give any further description of this 3-year-old product. We used three of

the functions in the program: Business Disk Winmark, High-End Disk Winmark and

measuring of access time.

The drives were formatted as logical NTFS units and the tests were run three

times. It’s the average value that is shown at the Winmark tests and the lowest

value on access time.

I think we should

ignore the values for the Sil3114 controller, as they aren’t correct. It seems

like the Raptors do have a slight problem with ICH5R. Anyhow, the Hitachi drive

is taking the lead on all controllers, even though the figures do not differ

that much between all the drives.

I’m starting to

suspect there is something wrong with the Sil3114 controller. The Samsung drive

increases the performance by about 20%, which isn’t the case on any of the other

controllers. The pattern is the same though, Raptor 74 GB goes off like a shot

followed by Raptor 36 and Hitachi, which seems to have a stable advance over

the other drives.

Next test is the

access time.

As you can see

it’s only Hitachi and Seagate that perform according to specifications. I think

we can pass the Raptor 36 as the measured value is very close to the specified

value. We will run an access time test with HDTach as well to see if the values

are coming back. Results and info on next page.

HDTach >

---

HDTach

HDTach is actually

an application that you shouldn’t use, as it is way too unreliable. We choose

to use it anyway because it is a popular and easy program, but the generated

results will not carry a lot of weight and should be taken with a pinch of salt

by all of you. The software itself doesn’t need any further presentation. The

partitions were removed to make the write test possible, and each test was

only run once per controller and drive.

As you all know

the result comes up as a graph and some values. We ended up with a lot of graphs

and we decided not to put them directly on the page. To see a specific graph,

please click the appropriate link in the table.

Harddrive	HDTach 2.61 (ICH5R)	HDTach 2.61 (Sil3114)	HDTach 2.61 (Promise TX4)	HDTach 2.70 (ICH5R)	HDTach 2.70 (Sil3114)	HDTach 2.70 (Promise TX4)
Hitachi Deskstar 7K250 160GB	Click me	Click me	Click me	Click me	Click me	Click me
Samsung Spinpoint P80 160GB	Click me	Click me	Click me	Click me	Click me	Click me
Seagate Barracuda 7200.7 160GB	Click me	Click me	Click me	Click me	Click me	Click me
WD Caviar SE 160GB	Click me	Click me	Click me	Click me	Click me	Click me
WD Raptor 36GB	Click me	Click me	Click me	Click me	Click me	Click me
WD Raptor 74GB	Click me	Click me	Click me	Click me	Click me	Click me

We start with CPU

Utilization tests in HDTach 2.61 and 2.70

It’s hard to find

a pattern here. The WD Caviar SE uses the least CPU power. The chip from

Silicon Image uses the least CPU power and ICH5R uses the most CPU.

You could almost

say the same thing here. On the other hand you can see there’s a tendency the

Raptor 74 GB needs the most power from the CPU. But everything depends on controllers.

A question of compatibility maybe? In general we come to one conclusion of these

tests: the slower drive the lesser power is required by the CPU. But that’s

not that surprising, really.

We continue with

the results from the Read Burst Speed test in HDTach 2.7

We choose not to

make a graph of these results when we used the Promise SATA150 TX4 controller.

Those of you that had a look at the graphs on top of this page maybe noticed

that the speed got a bit too high on a couple of drives (200+ Mbytes/s. 150

Mbytes/s is the top limit for SATA1.0 and 133 MBytes/s for PCI). Useless results

in other words.

I’m starting to get really impressed by the Hitachi drive, which again is just

a bit better than the other drives. The Samsung drive shows nice performance

as well.

We continue with

the last HDTach test: access time. Here’s the table from HDTach 2.61.

Harddrive	Access time specified (ms)	Measured access time ICH5R (ms)	Measured access time Sil3114 (ms)	Measured access time Promise SATA150 TX4 (ms)	Latency (ms)	Access time (Measured access time – latency)
Hitachi Deskstar 7K250 160GB	8,5	12,2	12,4	13,5	4,17	8,03
Samsung Spinpoint P80 160GB	8,9	13,2	14,8	14,9	4,17	9,03
Seagate Barracuda 7200.7 160GB	8,5	12,2	12,3	12,3	4,17	8,04
WD Caviar SE 160GB	8,9	13,9	14,4	14,1	4,2	9,7
WD Raptor 36GB	5,2	9	8,8	8,6	2,99	5,61
WD Raptor 74GB	4,5	8,2	7,9	8,2	2,99	4,91

Once again it is only the Hitachi and the Seagate drives that can keep up to

its specifications, even though Samsung is not far away from making it.

The results from

HDTach 2.7.

Harddrive	Access time specified (ms)	Measured access time ICH5R (ms)	Measured access time Sil3114 (ms)	Measured access time Promise SATA150 TX4 (ms)	Latency (ms)	Access time (Measured access time – latency)
Hitachi Deskstar 7K250 160GB	8,5	12,2	12,6	12,2	4,17	8,03
Samsung Spinpoint P80 160GB	8,9	13,3	15	14,7	4,17	9,13
Seagate Barracuda 7200.7 160GB	8,5	12,3	12,4	12,3	4,17	8,14
WD Caviar SE 160GB	8,9	14,1	14,1	14,1	4,2	9,9
WD Raptor 36GB	5,2	8,7	8,8	8,6	2,99	5,61
WD Raptor 74GB	4,5	8,1	7,9	8,2	2,99	4,91

Same thing once again. We told you that HDTach was a bit unreliable, but now

we’ve ended up in similar situations with two different versions and another

application. This makes you wonder, doesn’t it?

---

The final test we run

is from our colleagues at X-Bit Labs. Since the File-Copy-test in PCMark04 isn’t

quite reliable, we have chosen to use X-Bit Labs FC-Test, or File-Copy Test if

you like that better. This is a handy little program that has a couple of options

to measure read, write and copying speeds. First the program creates the files

you want to use, time is taken and with a simple calculation the speed is given.

It’s the same thing with read speed, files are read and time is taken and

one simple calculation later you get the read speed. Naturally is it the same

thing with copying. If you would like to know more about the program you read more here.

We will be using three different patterns that the program has at its service.

The three patterns are:

1. A big quantity

of smaller files, a typical Windows directory. (9006 files, 1060MB, average

per file:118kb)

2. A big quantity of bigger files, a typical MP3-directory. (271 files, 990MB,

average per file: 3.65MB)

3. A couple of big files, a typical ISO-directory. (3 files, 1600MB, average

per file: 533MB)

We will be doing two different copying tests, one that copies from one partition

to the same partition on the same disc (Copy Near) and one that copy from one

partition to another partition on the same disc (Copy Far). The drives were

split up to two identical equal size logical units and were formatted with NTFS.

The test was run three times per drive and as usual is it the average that we

show.

We start with the ISO tests, more precisely the write test.

The results are somewhat like each other on all the three controllers. 74GB

Raptor takes the lead but it is the Samsung drive that is the fastest 160GB

drive in this test. Once again the Silicon Image controller shows problems,

this time with the Hitachi drive. Let’s go on with the read test.

The results this time where very different. Raptor 74GB is way ahead of the

other drive this time as well.

The Hitachi drive is the on that show the best average result even if the Samsung

drive worked extremely well with the Promise controller. All drives perform

well in this test, with the exception of the reference drive, Raptor 36, which

totally fails this test. On to the first copy test, where we copy from and to

the same partition.

All I can say is Samsung! Samsung’s drive was the one which had the highest

specified internal transfer rate among today’s 160GB-drives and it really

shows this here. 74GB Raptor takes the lead only in the last test. But the big

question is what happened to the 36GB Raptor?

The Silicon Image controller has given us a lot of strange results during this

entire review and it keeps doing it here. All drives perform considerably worse

with that controller. The 74GB Raptors results speak for itself there. Now its

time for the copy test were we copy from one half of a drive to another.

The results are similar to the last test. Once again we ignore the Sil3114,

because today we are testing the harddrives, not the controllers. That’s

why we won’t care about this controller results anymore. The Raptor 74GB

and the Samsung drive shows what its made of here, and the Raptor 36GB isn’t

too far behind.

---

FC-Test – MP3

Let’s go on with the MP3-tests. In this test, as described before, there

are many files that shall be written, read and copied, all in one sweep. We

start with the write tests.

Samsung really impress in these tests. The 74GB Raptor takes the total lead

with a very good result with the Promise controller. Otherwise the results are

similar to those in the ISO write test.

The Raptor 74GB and Hitachi is fighting for the lead here and it’s Hitachi

that takes it in this test. Also Samsung and Seagate perform well in this test.

The question is what’s wrong with the 36GB Raptor which really doesn’t

perform well in these tests.

The copy king Samsung

strikes again, closely followed by Hitachi.

In this test the Samsung does not keep up as well as in the other copy tests,

and is replaced by Hitachi, followed by Seagate and Samsung. The Raptor 74GB

shows what it’s made of on the Promise controller where even the 36GB

does good for once. One thing that is sure is that the WD 160GB is the slowest

drive so far.

---

FC-Test – Win

Here are the final

tests in this review – the Windows tests. Here we have a really large amount

of small files that will be written, read and copied.
A decisive factor

in these tests is if the disks have fast or slow electronics. Rotation speeds

and other properties doesn’t do a large impact here.

Samsung is once

again very strong in these non-synthetic tests. The 74 GB Raptor wins this round

as well on the Promise controller. They seem to have good compatibility with each other.

Other than that, Samsung takes the lead, followed by Hitachi and then the two

others – Seagate and WD 160GB.

The Hitachi performs

the best, but it’s still Seagate here that shows a strong side. I suspect that

this is because of its native SATA host.

The Marvel bridge on the other disks decreases the performance by a minimum

in ordinary cases, but many a little makes a mickle.

Once again WD 160GB takes the bottom spot.

It’s copying time

again and Hitachi and Samsung performs well as usual.

Once again, Hitachi

takes the lead followed by Samsung. Once again WD 160GB is at the bottom.

For summing up the entire

FC test we can say this – The 74 GB Raptor is the best on most of the tasks,

but when it comes to 160 GB it’s Hitachi and Samsung that really performs in

addition to the others. WD Caviar SE is the test’s big loser that placed itself

on the last spot in nearly every test. The Raptor 36GB really didn’t suffice

either.

---

General

With this page

we wanted to present a couple of aspects that has nothing to do with performance.

We’ll start with

a price comparison.

Harddrive	Price (SEK)	Effective storage space (MB)	Storage per SEK (MB/SEK)
Hitachi Deskstar 7K250 160GB	1050	157065	149,6
Samsung Spinpoint P80 160GB	1100	152625	138,8
Seagate Barracuda 7200.7 160GB	1100	152625	138,8
WD Caviar SE 160GB	1150	152625	132,7
WD Raptor 74GB	2200	70904	32,23
WD Raptor 36GB	1100	35299	32,09

First of all we

can remove the two raptors if we’re after a disk with a lot of storage space.

You can see in the table above that there aren’t any big differences in

price in this sector, but Hitachi gives you about 4500 MB more storage space

than the competitors.

At the same time as the Hitachi belongs to the cheapest ones we have a great

value on the amount of MB/SEK.

If we move on to

another aspect many are thinking about it’s the noise level on the drives. Since

we don’t have a sophisticated method of analyzing of sound levels we have to

use the writer’s ear and what I can tell you is that all the disks are very

quiet, except for the two Raptors that can make a lot of noise sometimes. To

choose a winner in the war of silence it’s Seagate that wins. It’s their motto,

so it wasn’t really unexpected.

The Seagate is barely audible even at full load.

The next aspect

is the disk temperature. These measurements are made with SpeedFan via S.M.A.R.T.

so take these results with a pinch of salt. It’s really the surface temperature

that is the most interesting (the energy being released) but these values can

be seen as some form of hint. We’re planning to acquire a real thermometer for

the test lab so that we can measure surface temperatures in the future. So,

these are the temperatures inside the disks after some time of load.

Harddrive	Temperature (°C)
Seagate Barracuda 7200.7 160GB	51
Samsung Spinpoint P80 160GB	52
WD Caviar SE 160GB	53
Hitachi Deskstar 7K250 160GB	54
WD Raptor 36GB	54
WD Raptor 74GB	55

Our thought from

the beginning was to include an aspect rarely seen in reviews – return frequencies

on the disks. Unfortunately we’re yet to succeed in collecting this information.

We hope that we have better luck next time.

---

Conclusion

Finally time for the last page in this roundup: the conclusion.

To start with we

can say that Western Digital Raptor 74 GB is the king among today’s ATA harddrives.

It performs the best in almost every test and beats its predecessor, Raptor

36 GB, with good margins in most cases. The Raptors are not regular drives made

for any home user, but if you are a person in hurry who would like to load Windows

or games a few seconds faster, this could be the way. These drives are made

for professionals and users who need good performance in general. Systems

where these drives really could be handy are servers and as well for machines

used for image processing and movie editing on a large scale. Personally I would

use RAID for such a system, as a single drive wouldn’t improve the performance

enough. A RAID0+1 setup with Raptor 74 GB wouldn’t be too bad.

When it comes to

the 160 GB S-ATA drives, which was the main purpose with this review, we have

an obvious winner and that is Hitachi Deskstar 7K250. It’s got the best performance,

sometimes with a large margin, in most tests. The drive also gives the best

value for the money, as it is a little bit larger than the other drives. Samsung

SpinPoint P80 takes the second place in the performance race. It performed very

well in the synthetic copy test. We add bonus points for the design. It may

seem a bit ridiculous, but it was very easy to grip the drive by the two hollows

during installation and such. If you want a drive as quiet and cool as possible

it’s still Seagate you should go for. The big loser in this roundup is Western

Digital Caviar SE that has the worst performance in almost all the tests.

As we tested all

the drives on three different controllers we should add something about them

as well. Our expectations before the tests was that the integrated controller

in the mainboards north bridge, ICH5R, would perform the best as the other two

runs through the PCI bus, but it wasn’t that simple. Promise Technology TX4

showed best performance in most tests and that maybe isn’t that strange when

it come to single drives. Promise is, unlike Intel; a company specialized on

controllers. The ICH5R will probably not show its advantages until you run with

RAID, where the PCI bus will become a bottleneck. The PCI chip controller from

Silicon Image had a number of problems in the tests today and because of that

we chose to exclude it from the test system.

NordicHardware >