Pentium III 500 Platform Comparison

Pages: 1 2 3 4 5 6 7 8

Application Benchmark Performance and Conclusions

While all of the component level test might be interesting, and provide some insight into the relative strengths and weaknesses of each platform, what it really comes down to for the vast majority of people is how their applications will perform. Technically, the only way to know for sure how your applications will run is to test them under the exact conditions they will be used in the ‘real world’, however, that is obviously not feasible for a comparison such as this.

For my purposes, I decided to use the ZDBop benchmarks for the system level testing because they are readily available to anyone wanting to verify the results given here, and because they are generally considered the best Windows based application benchmarks. There are other benchmarks, such as those offered by BAPCo, however they are not as readily available to the average user, and they lack some of the basic business applications that the Winstone99 business tests include. Perhaps future platform comparisons will include Sysmark test results as well.

Benchmark

AX6C/DRDRAM

AX6BC/SDRAM

AX63/SDRAM

AX63/VCSDRAM

Winstone99 Business

27.5

28

27.8

27.5

Winstone99 HighEnd

23.9

24.6

24

23.9

…AVS Express 3.4

4.66

4.75

4.61

4.61

…Front Page 98

1.71

1.78

1.76

1.74

…MicrostationSE

2.63

2.64

2.61

2.63

…Photoshop 4.0

3.34

3.42

3.32

3.42

…Premiere 4.2

2.38

2.42

2.43

2.44

…Sound Forge 4.0

1.63

1.69

1.57

1.56

…Visual C++ 5.0

2.37

2.52

2.51

2.42

Content Creation 2000

18.8

20.1

19.5

19.4

Passmark Overall

64.2

63.0

67.1

67.9

We can clearly see from these results why the BX chipset is still considered the ‘performance’ chipset on the market. Though the differences are not huge, Winstone99 seems to favor the BX chipset in almost every test. What is also very interesting is that the two applications considered the most memory intensive (Photoshop and Premier) perform better with the Pro133 chipset using VCSDRAM.

The question here is whether these results show the actual performance differences that users will experience in the ‘real world’. Unfortunately, ZDBOp does not provide their weightings of the various tests they perform. As we saw in the various component level benchmarks, the BX chipset seems to be better with graphics and CPU performance, while the Pro133 has better I/O performance. Memory performance is very close, and might be called a draw.

So, is this reflected in the Winstone99 test scores? As an illustration and counterpoint, I’ve included the Passmark overall test results at the end of the table. Though it is not an application level benchmark, I thought it would be interesting to show how Passmark interprets the individual results into an overall score. It would seem that Passmark gives a greater weight in their total to I/O performance, while Winstone99 favors either graphics, CPU or both.

In the final analysis, most people will believe what they want to believe, and most believe that the BX chipset has better memory performance. As shown here, that is not necessarily true, as much depends upon how the application actually uses memory. It is also apparent that the performance difference is really very small between all of these platforms in most business applications, so the real issue becomes on of price and reliability. For those who have specialized requirements, such as graphics, processor or I/O intensive applications, it might be wise to look at the results of the various component level tests and choose based upon those results. While the system level benchmark results are the ones most commonly reported, they may not reflect the way that you use your system.


Pages: « Prev   1 2 3 4 5 6 7 8   Next »

Be the first to discuss this article!