So Microsoft is walking on air now that they are legally free and clear to ship Windows 98 next month (though their 500+ lawyers will be working overtime to prepare for the big trial in September). Predictions of Windows 98's sales figures look good, but will that change now that CNET's done some preliminary performance testing of the new OS? They set out to answer the question, "Is Windows 98 faster than Windows 95?" (Never mind that since it's an upgrade, the question shouldn't be "if," but "how much?") The bad news is, Wintel users expecting a speed increase may be sorely disappointed.

According to CNET's initial testing, Windows 98 degrades performance by up to 23 percent on a lower-end system. Performance loss was not as severe on high-end machines-- only 6 percent-- but that's still a reduction. However, Windows 98 improves performance on systems that are currently running Windows 95 and Internet Explorer 4. (The very fact that installing Internet Explorer 4 on your Windows 95 system slows everything down should form the basis for an excellent Netscape ad-- if they ever actually advertised their browser, that is.) Note that Windows 98 apparently includes an admittedly cool dynamic and gradual defragmentation function, which should improve the performance of certain disk-based activities (such as application launching) over time. So it may turn out that, after a few weeks, Windows 98 is faster after all, but we're not holding our breath.

It never ceases to amaze us that Microsoft continues to ship "upgrades" that actually slow things down. Actually, that's not entirely accurate; what we mean to say is, we're constantly perplexed that people continue to buy these upgrades. In contrast, the Mac OS has improved performance on PowerPC systems in each release since 7.6, and that trend reportedly continues with Allegro (Mac OS 8.5), now expected this fall. Yet another benefit of using a Mac: system upgrades that are actually upgrades. That's not even innovation-- that's just common sense.