iXBT Labs - Computer Hardware in Detail

Platform

Video

Multimedia

Mobile

Other

Testing of hard disks with interface IDE in July 2000

September 4, 2000



This year the manufacturers of hard disks please us with novelties even too regularly. It would be desirable to cover everything but we have to limit ourselves with what we have at the moment until some of disks become out of date.

So this is what we have from corporations-manufacturers.

Fujitsu

Series Fujitsu MPF3xxxAT

Bit density - 7.8 (7.8 GBytes/in2 or 10.2 GBytes on a plate). Rotation speed - 5400 rev/min. It is positioned as optimum for computers of initial level from the point of view of entertaining and business - applications and also digital video processing and reception HDTV. The company underlines high speed of large size files access. Announced maximum linear read rate is 37.8 MBytes/sec. Among technological decisions it is possible to mark the following:

  • Heads with huge GMR of the 4th generation
  • Advanced construction of the upper cover permitted to decrease noise up to 3.4 bel

Besides as always guaranty No Excuses « is paid special attention to. The set contains disks of capacity 10.2 (MPF3102AT), 15.3 (MPF3153AT) and 20.4 (MPF3204AT) GBytes. We had the model of capacity 15.3 GBytes. We hope to add the model of series MPF3xxxAH (with rotation speed 7200 rev/min).

IBM

It is a representative of the series Deskstar 75GXP (IBM DTLA 7200 rpm)

Here there is already a qualitative difference from the previous series - glass media, differential preamplifier and heads with huge GMR of the 5th generation have allowed to magnify bit density on the track on about 40%. Besides (as far as I know for the first time on disks with interface IDE) the thermal monitor has appeared. One more novelty is IBM safety ramp - the device lifting the head above the surface of the disk after deenergization - practically completely (as the technicians of the company suppose) eliminates the possibility of information loss at transportation of the disk. And if to take into account a new ceramic spindle and a laminated upper cover permitting to reduce the noise level up to 7 dB and at the same time not changing the habitual design, then it is possible to say that even before installation into the computer the disk impresses. Naturally all previous technological decisions such as TrueTrack servo, NoID sector formatting and Drive Fitness Test are remained. The range of capacities - from 15 up to 75 GBytes satisfies any taste. We had "the average" model of capacity 45 GBytes. As for the "paired" series Deskstar 40GV (DTLA 4500 rpm) we also hope to add it.

Quantum

In this company the leader is Fireball Plus LM

There is a common standard set of decisions for modern disks there - bit density 10.2 GBytes on a plate, firm technologies of data protection - Quantum Data Protection System (DPS) and protection against impacts Shock Protection System (SPS). Variations in design are not noted. The set contains disks of capacity 10.2 GBytes, 15.0 GBytes, 20.5 GBytes and 30.0 GBytes. We tested the disk of capacity 20.5 GBytes

Seagate

Here we have two leaders - in the category 5400 rpm it is the model U10 and in the category 7200 rpm it is Barracuda ATA II.

As for the series U10 the first thing the disks attract our notice by is a bit unusual design. The protective cover made of soft plastic SeaShield closing almost all electronics engineering and intended to fence it from the "effect" of user's hands. Besides there is a box made of transparent plastic where the disk is placed. It is necessary to say that they look impressive. In this series the protection system from impacts G-Force Protection and also the system of diagnostics 3D Defense System are used. The bit density is 10.2 GBytes on a plate. The pride of the company is the install utility Disk Wizard. Besides as usual noise reduction up to 3.2 bel is paid special attention to. The bar contains the disks of capacity 20.4, 13 and 10.2 GBytes. We tested the high model.

To my mind the release of Barracuda ATA II was waited for. At least the previous model was interesting enough. In comparison with Barracuda ATA the size of the buffer was increased up to 2 ╠byte and it should eliminate one of "narrow places" and naturally the bit density - up to "the world standard" 10.2 GBytes on a plate. 3D Defense System and Disk Wizard are present too. The set contains disks of capacity 30.6, 20.4, 15.3 and 10.2 GBytes. We got the largest model more than 30 GBytes.

Western Digital

This company is represented by the bar Caviar AA and BA and also by the disk for home entertainments WD Performer - of the latest fashion. We shall start with it.

So it is WD Performer 272AW of the capacity accordingly 27.2 GBytes. We look at it and see ů the old disk WD! Since those times the IBM parts were applied only in the disks by IBM. Let's look what the company says about this set. First that pleases is that there are no data sheets, is not pointed even what models the bar consists of. But it is explained in detail how good it is to watch TV having digital recorder with a built - in hard disk and also how useful is the hard disk for listening music. Then it is explained that home entertainment disk is not the computer disk. It should provide the constant high speed of data transmission, support multithreading well (for example to watch one film and to record another one) and also to be practically silent not to deafen music. We are promised that Performer satisfies all conditions as the firm technology WhisperDrive is used in it. Then let us check. Besides it is affirmed that capacity of home entertainment disks should be measured not in gigabytes but in hours of playback time. It is not clear only why their disk was marked all the same in gigabytes.

Then there is the series Caviar AA with the spindle rotation rate 5400 rpm. And here is the same picture! As for the remaining everything is usual - bit density 10.2 GBytes on a plate, technology of data protection Data Lifeguard. There is a problem with the determination of the bar. As all disks are marked in the same way - by capacity and we can get 20.5 GBytes as 2 plates with 10.2 and as 3 ones with 6.8 then we had to look through the data sheets.

In my opinion the new bar contains disks of capacity 10.2, 15.3 and 30.7 GBytes (it would be logical to release and 20.4 but there is only 20.5 and it has three plates). There is a disk and in 45 GBytes but it has 3 plates, that is the bit density is already 15 GBytes on a plate but it is another story. We had 30.7 GBytes disk.

There is the lack of the new series BA but as it was already said if we waited for everything, the review would appear just to the moment the disks became out of date. We are already a bit late. At last before the tests we shall give the summary table of the disks characteristics.

Fujitsu MPF3xxxAT
IBM DTLA-307075, 307060, 307045, 307030, 307020, 307015
Quantum Fireball Plus LM
Seagate U10 ST320423A, ST315323A, ST310212A
Seagate Barracuda ATA II ST330630A, ST320420A, ST315320A, ST310210A
WD Caviar WD307AA, WD153AA, WD102AA
Capacities 20.48, 15.36, 10.24 GBytes 75, 60, 45, 30, 20, 15 GBytes 30.02, 20.55, 15.02, 10.27 GBytes 20.4, 15.3, 10.2 GBytes 30.6, 20.4, 5.3, 10.2 GBytes 30.7, 15.3, 10.2 GBytes
RPM 5400 7200 7200 5400 7200 5400
Buffer Size 512 KB 2 MBytes(*) 2 MBytes 512 KB 2 MBytes 2 MBytes
Heads 4, 3, 2 10, 8, 6, 4, 3, 2 6, 4, 3, 2 4, 3, 2 6, 4, 3, 2 6, 3, 2
Plates 2, 2, 1 5, 4, 3, 2, 2, 1 3, 2, 2, 1 2, 2, 1 3, 2, 2, 1 3, 2, 1
Average Latency 5.56 ms 4.17 ms 4.17 ms 5.56 ms 4.16 ms 5.5 ms
Average Seek Time 9.5 ms 8.5 ms 8.5 ms 8.9 ms 8.2 ms 9.5 ms
Average Track to Track Time 1.2 ms 1.2 ms 0.8 ms ? ? 2.0 ms
Full Track Time 18.0 ms 15.0 ms 15.0 ms ? ? 19.0 ms
(*) Upper 132 KBytes are occupied by firmware.

The tests

The testing was conducted on the following system:

  • System board Chaintech 6BTM (Intel 440BX BIOS);
  • Controller Ultra ATA/66 Abit HotRod66 (HighPoint HPT 366 BIOS);
  • Processor Intel Pentium II 400 MHz;
  • 128 MBytes SDRAM SEC PC100;
  • Videocard ASUS V2740 (Intel 740);
  • Operating system Windows 98 SE

Since the given version BIOS the controller HighPoint support the choice of the load device thus it is not necessary to be loaded with primary master. I advise everyone who has not made it yet to flash new BIOS.

The standard preamble :-). Contents of all tested disks were identical. Cloning was effected with the help of the utility Norton Ghost. The standard set of the tests - WinBench 99 (Business WinMark and Disk Inspection Tests) and Adaptec ThreadMark 2.0 was used. The test Disk CPU Utilization can be carried out in two modes: with the fixed read rate and with maximum accessible. The testing was conducted at the fixed rate of 4000000 bytes/sec. The test WinBench 99 was repeated three times, the test Adaptec ThreadMark 2.0 - five times. In the tables the average values are given. At testing the support DMA in Windows was included. The tests were conducted on "cold" disks that is before realisation of each following test, the time for cooling up to room temperature was given to the disk. Disks before realisation of each test were defragmented with usage of the standard utility Defrag included in the package Windows 98 SE. At testing the disks were connected as master to the first channel of the controller HotRod.

So for the beginning we shall have WinBench 99. The disks are grouped in the following way - at first 7200, then 5400.

IBM DTLA-307045
Quantum Fireball Plus LM
Seagate Barracuda ATA II ST330630A
Fujitsu MPF3153AT
Seagate U10 ST320423A
WD Caviar WD 307AA
WD Performer 272AW
Business Disk WinMark 99 (thousand bytes/sec)
5750 (65.68%)
4680 (53.41%)
5130 (59.45%)
4030 (46.65%)
4020 (45.97%)
4600 (52.26%)
4390 (50.47%)
High-End Disk WinMark 99 (tb/s)
18300 (42.18%)
14800 (35.28%)
15200 (35.27%)
12600 (28.54%)
11800 (27.75%)
12700 (31.22%)
11900 (29.59%)
Disk Transfer Rate: Beginning (tb/s)
37400 (16.5%)
26700 (11.4%)
29700 (7.52%)
23800 (8.62%)
23700 (8.9%)
15900 (5.95%)
19100 (4.89%)
Disk Transfer Rate: End (tb/s)
17600 (16.5%)
19500 (11.4%)
19900 (7.52%)
14100 (8.62%)
15200 (8.9%)
14400 (5.95%)
10800 (4.89%)
Disk Access Time (ms)
13 (1.38%)
12.7 (2.34%)
13.7 (2.24%)
14.9 (1.11%)
18.3 (0.93%)
15.6 (1.08%)
16.4 (1.01%)
Disk CPU Utilization (Percent Used)
7.97
7.72
5.32
7.42
7.55
7.53
7.36
High-End Disk WinMark 99 in depth
Disk Playback/HE: AVS/Express 3.4 (tb/s)
15200
11900
12100
10200
8410
8720
7950
Disk Playback/HE: FrontPage 98 (tb/s)
37300
36100
36700
33600
32400
35400
34800
Disk Playback/HE: MicroStation SE (tb/s)
17600
14700
15100
13300
12800
12900
12000
Disk Playback/HE: Photoshop 4.0 (tb/s)
12200
9740
9880
7490
7410
8720
7650
Disk Playback/HE: Premiere 4.2 (tb/s)
17200
12900
13400
10900
10600
11400
10500
Disk Playback/HE: Sound Forge 4.0 (tb/s)
31600
24500
23900
20400
20300
21700
21400
Disk Playback/HE: Visual C++ 5.0 (tb/s)
17300
14300
15200
12200
11700
14000
13400

Well for IBM there are still no competitors. Besides other things it is possible to mark access time to DTLA. If to add the stated value of search time to the delay factor, we shall get the digit exceeding the obtained result. It is not bad for such "not small" disk! Barracuda practically everywhere is slightly faster than Quantum. The exception is only Sound Forge. Besides it is possible to mark that Barracuda loads the processor at linear reading very little.

Then about disks with a rotation speed 5400 rpm. First it seems that Fujitsu and U10 are twins. But look at the access time for U10! Softly speaking it does not go with the stated digits. Probably the lagging in HighEnd Winmark is from here. As for the disks WD then it is a special talk. By integrated evaluations of the tests Caviar AA is a leader and Performer is quite good. But look at exchange rate. Obviously something is wrong. To make it clear let us look at the linear reading graph.

IBM

After the previous review I was told that IBM couldn't have such graphs but I got only these ones. I tried to put this disk on the antivibrational seat

and rigidly fixed it in the regular internal 3-inch bay of InWinSolist case.

But nothing helped and the graph all the same was curve. Not to spoil the reputation of the respected company I shall give the graph from Storagereview (http://www.storagereview.com/). It should be like this:

Look at the graph Barracuda or compare the graphs U10 here and on Storagereview and you can see that such curve graph was got only for IBM.Then here is the illustration of usage IBM DTLA disk with the controller ATA 33:

It is rather curve too but at the same time instructive.

Quantum

The graph is normal:

In my point of view small-sized ripple is explained by operation of algorithms " reading forwards "

Seagate Barracuda ATA II

We can do it!

The graph is good! It was interesting to know if the characteristic bend was remained at operation in ATA 33 mode. And it was!

We still don't have the real explanations to this effect.

Now about the disks with the rotation speed 5400 rpm.

Fujitsu MPF3xxxAT

The graph is normal.

Seagate U10

It is almost exact repetition of the graph with storagereview:-)

But there is one more question. Does only Barracuda have this bend or is it a property of any disk Seagate with interface IDE. As experience shows:

It is only Barracuda's quality.

Western Digital

I do not separate series on purpose as Caviar AA

And Performer:

Both behave themselves the same way darkly. The connection of the disk Caviar AA to the controller ATA 33 has resulted only to the displacement of spikes:

At the moment the investigation of this case is being carried on together with the staff of the company ELKO gave disks. The results were also confirmed on the test benches of the company therefore the main version now is that algorithms of disks caching and the algorithm of reading the test don't match. Any way information will be added after any results (positive or negative for disks) are got.

Next in turn is Adaptec ThreadMark

In the first column of the table given below the average weighted value of exchange rate in MBytes/sec as the results of 64 tests is indicated. During these tests reading and record operations were made by blocks at 512 byte, 1, 2, 4, 8, 16, 32 and 64 kilobyte by one, two, three or four thread at the same time with usage of standard calls Win32 API . In the second column you can see the processor loading at tests.

HDD
Below the average weighted
value of exchange rate in Mbytes/sec
processor loading
at tests
IBM DTLA-307045
15.48
53.52%
Quantum Fireball Plus LM
10.97
42.61%
Seagate Barracuda ATA II ST330630A
12.61
47.92%
Fujitsu MPF3153AT
9.42
39.14%
Seagate U10 ST320423A
10.14
42.38%
WD Caviar WD 307AA
12.58
51.38%
WD Performer 272AW
11.19
50.34%

Here we have something interesting. As for IBM the competitors still are not present but WD fulfils its promises about good support of multistreaming. By the averaged value the disks of this company are practically at the level of the disks with the speed 7200 rpm.

Temperature characteristics

You can blame me but I shall not give the digits because the testing lasted too long and the weather changed too often. I can only said that none of the models was heated appreciably more than its predecessor. And naturally disks on 7200 are "hotter" than disks on 5400.

Acoustic properties

I say once again that all modern disks are silent. In my opinion the most silent are from Fujitsu and WD, the loudest is from Quantum.

Summary

Everything is clear with the leader - it is IBM. The second Barracuda did its best to disposed of some disadvantages of its predecessor but not having lost its advantages. As for Fireball Plus LM - it appeared first and waited for the worthy competitors for rather long time and it is its alibi. As for disks with the rotation speed 5400 rpm I shall not make the final conclusion before such strange behaviour of disks WD is cleared up. Serious work is being done in this direction. In my opinion from Fujitsu MPF3xxxAT and Seagate U10 the winner is all the same Fujitsu.

The conclusion

After each review I naturally receive a lot of the letters with criticism. Some people think that some kinds of tests are not necessary and so on. Because of this it would be desirable to begin the discussion on pages of our conference on the topic: "What hard disks tests we need". On the same storagereview under Windows 2000 with NTFS. Is it necessary for us? It is possible to do testing by any tests if it would be interesting for the majority of people. I am waiting for constructive suggestions.


Write a comment below. No registration needed!


Article navigation:



blog comments powered by Disqus

  Most Popular Reviews More    RSS  

AMD Phenom II X4 955, Phenom II X4 960T, Phenom II X6 1075T, and Intel Pentium G2120, Core i3-3220, Core i5-3330 Processors

Comparing old, cheap solutions from AMD with new, budget offerings from Intel.
February 1, 2013 · Processor Roundups

Inno3D GeForce GTX 670 iChill, Inno3D GeForce GTX 660 Ti Graphics Cards

A couple of mid-range adapters with original cooling systems.
January 30, 2013 · Video cards: NVIDIA GPUs

Creative Sound Blaster X-Fi Surround 5.1

An external X-Fi solution in tests.
September 9, 2008 · Sound Cards

AMD FX-8350 Processor

The first worthwhile Piledriver CPU.
September 11, 2012 · Processors: AMD

Consumed Power, Energy Consumption: Ivy Bridge vs. Sandy Bridge

Trying out the new method.
September 18, 2012 · Processors: Intel
  Latest Reviews More    RSS  

i3DSpeed, September 2013

Retested all graphics cards with the new drivers.
Oct 18, 2013 · 3Digests

i3DSpeed, August 2013

Added new benchmarks: BioShock Infinite and Metro: Last Light.
Sep 06, 2013 · 3Digests

i3DSpeed, July 2013

Added the test results of NVIDIA GeForce GTX 760 and AMD Radeon HD 7730.
Aug 05, 2013 · 3Digests

Gainward GeForce GTX 650 Ti BOOST 2GB Golden Sample Graphics Card

An excellent hybrid of GeForce GTX 650 Ti and GeForce GTX 660.
Jun 24, 2013 · Video cards: NVIDIA GPUs

i3DSpeed, May 2013

Added the test results of NVIDIA GeForce GTX 770/780.
Jun 03, 2013 · 3Digests
  Latest News More    RSS  

Platform  ·  Video  ·  Multimedia  ·  Mobile  ·  Other  ||  About us & Privacy policy  ·  Twitter  ·  Facebook


Copyright © Byrds Research & Publishing, Ltd., 1997–2011. All rights reserved.