ATA/100 – Real Performance or Marketing Hype?

Pages: 1 2

Test Setup and Results

OK, you say ‘… prove it…’, so here you go. I used more then one chipset for the tests, just to make sure there weren’t any chipset related issues. Also, since some are convinced VIA chipsets have performance bottlenecks that limit benchmark results, I used an Intel 815E based motherboard as a baseline. I also thought of using two different brands of hard disk, but many hours of experience has shown me that there is no difference (mostly with WD, IBM and Maxtor) between different brands.

<b>The Test Setup</b>

Mainboard

Soyo 7VDA

Soyo 7ISA+

Chipset

VIA Apollo Pro 266

Intel 815E

CPU

Intel Pentium III 800EB

Same

Memory

128MB PC2100 DDR SDRAM

128MB PC133 CAS 2

Hard Disk

WD 200BB – 20GB ATA/100 7200 RPM

Same

OS

Win98 SE

Same

Busmastering Driver used

MS Win 98 SE driver

Intel – supplied with 815 .inf update

Win HD cache

Auto

Same

Video card

Cardex GeForce II MX 32MB

Same

Display mode

1024×768-64K @ 85Hz

Same

Why did I choose these two Mainboards? Simple, they both have the BIOS fully developed to give full ATA/100 burst speeds. All Mainboards I’ve tested that have the VIA 686B (with ATA/100 rating) Southbridge, still do not have the BIOS optimized for full ATA/100 burst speeds. This includes the Soyo K7VTA-B, K7VMM-B 7VCA2; MSI K7T Pro 2A and Gigabyte 7DX (AMD 761 Northbridge &amp; VIA 686B Southbridge). All of these burst in the 75MB/s range instead of the 85MB/s I would expect. The 7VDA and its Apollo Pro 266 DDR chipset use a new VIA Southbridge with ATA/100 support, while the ‘E’ in the 815 is for the new (enhanced?) Southbridge (or ‘I/O Hub’) with ATA/100 support.

Two totally different benchmarks were used. The first is HD Tach, since it is popular and reports the burst transfer (only way to tell which ATA mode is truly being used). The other is ZD Labs Winbench and Winstone, since some of their tests are application based and will give real world results, and because they are quite good and repeatable. One golden rule is never to trust just one benchmark, if you use two and they show different trends, then use a third… otherwise your results could be biased.

Now for the ‘hard numbers’. I ran each test 3 times, if the results from these runs were not within 3% of each other, I ran them 3 times again until they were within 3%. I then used the best score out of the 3. I also tested the VIA based mainboard with the standard Microsoft Win98SE VIA Busmaster drivers, the VIA 4-in1 4.26a pack and the 3.0.11 Busmaster only drivers. There was no gain over the Win98SE drivers, so those were the ones reported. I also tried the Intel Ultra ATA drivers over the standard ones supplied with the 815 .inf update, again with no gain so the standard ones were used. The hard disk was defragged before each run. To test in different ATA modes I used WD’s utility that allows setting the drive to ATA/33, 66 or 100.

Mainboard

7VDA

7VDA

7VDA

7ISA+

ATA mode

33

66

100

100

&nbsp; &nbsp; &nbsp; &nbsp; &nbsp;

HD Tach 2.60

&nbsp; &nbsp; &nbsp; &nbsp;

Random Access

10.9

10.9

10.9

11.4

Read Burst

30.6

59.5

85.3

85.7

Read Max.

24833

33097

32938

32932

Read Min.

18740

23189

22687

23088

Read Avg.

24180

25694

25940

25969

CPU Utilization

8.3

9.6

8.1

6.6

&nbsp; &nbsp; &nbsp; &nbsp; &nbsp;

Business Winstone 2001

34.5

35.4

35.6

34.8

Content Creation Winstone 2001

33.6

34.9

34.7

35.5

&nbsp; &nbsp; &nbsp; &nbsp; &nbsp;

Winbench 99

&nbsp; &nbsp; &nbsp; &nbsp;

Business Disk

6010

6080

6350

6280

High Disk

18100

19100

19300

19100

Disk Transfer – Beginning

24200

32200

32300

32400

Disk Transfer – End

24400

32900

32900

33000

Access

11.0

11.0

11.0

11.0

CPU Utilization

1.42

1.49

1.49

1.77

I’ll let the numbers speak for them selves, but as you can see the three ATA/66 and ATA/100 scores are pretty much the same and show just a slight gain over the ATA/33. Based upon these results, I see no reason to dump your ‘old’ ATA/66 mainboard for a new ATA/100, or even adding a PCI ATA/100 controller just to get more performance, because it just isn’t there. However, there is an advantage to using a newer 7200-RPM hard disk over an older 5400-RPM unit.


Pages: « Prev  1 2  

Be the first to discuss this article!