The Windows NT Magazine Lab used BAPCo's SYSmark/32 benchmark suite for Windows NT to test the performance of all 10 laptops in our roundup. We administered battery tests with power management turned off and on. We noticed little difference between the two power management settings. However, adjusting video drivers and screen resolutions affected the SYSmark/32 test score quite a bit in some cases: For example, the IBM ThinkPad 760ED jumped from seventh place to first place when we tested it at 800*600, the same resolution as all the other machines, instead of its maximum resolution (1024*768). We also tested compatibility with three PC Cards, which worked without a hitch on all the notebooks.
Rather than using a benchmark tool that generates a score based on a proprietary, synthetic test sequence, the Lab likes to test machines on software that you use every day. BAPCo's SYSmark/32 1.0 benchmark for Windows NT tests performance via the eight real-world applications listed in Table A.
The SYSmark/32 benchmark score is based on the amount of time the applications take to execute automated scripts performing the same tasks you might. The test runs each script four times and compares the average of these runtimes to a baseline system. A system that receives a score of 200 is twice as fast as one that rates 100. For more information about the BAPCo SYSmark/32 benchmark suite, visit BAPCo's Web site at http://www.bapco.com.
|The device used to test|
the batteries on the laptops.
To test the batteries, the Lab planned to turn on the notebooks and time how long it took them to run out of juice. Unfortunately, this procedure is not as easy as it seems: All the laptops have power-saving capabilities; they shut down after a few minutes (or even seconds) of keyboard inactivity. Would someone have to sit for hours at each notebook and press a key every few seconds?
The Lab investigated and found a few companies that build devices for pressing keys automatically, but we thought building something that would press keys might be fun. After all, how tough could it be? We came up with lots of ideas, but decided to use a bar hooked up to a slowly turning electric motor to depress a key.
We assembled our battery tester and tried it out. It worked fine, but we needed a way (without using the battery) to time how long each laptop lasted on battery power before shutting off. We designed and built a circuit that measured how long the screen stayed active. A photocell (or photo resistor) watched the screen and stopped timing when the computer and the display shut off. We recorded battery life in hours and minutes.
|TABLE A: The BAPCo SYSmark/32 for Windows NT Application Test Suite|
|Type of Application||Applications|
| Microsoft Word 7.0, Lotus WordPro 96
Microsoft Excel 7.0
Borland Paradox 7.0
Lotus Freelance Graphics 96, Microsoft PowerPoint 7.0
Adobe PageMaker 6.0