X
Tech

Is Vista really slow to start up?

According to a handful of guys on the Internet, Vista is slower than a Commodore 64 to start up. Oh really? On my test machine, I needed a stopwatch to measure the difference in startup times. So what's the real problem?
Written by Ed Bott, Senior Contributing Editor

[Update 16–Apr: See my follow-up post, More details about Vista’s startup times, for additional test results.]

Got a deadline and a news hole to fill? Why not head out on the Internets and do some research by digging up random posts on a message board or a newsgroup?

That’s apparently what Computerworld’s Gregg Keizer did earlier this week, with a piece lamenting the fact that Vista is slower than XP at start-up and shutdown. (I found it because my colleague Mary Jo Foley noticed it.) The evidence? Testimonials from a bunch of guys on the Internets, complete with inflammatory quotes and a requisite helping of snark. ("I've compared it to a Commodore 64 loading programs from tape, but I think the Commodore was faster." Heh.)

Now, I find it a bit puzzling that someone with all the resources of the IDG empire behind them can’t find some lab rats to actually run some tests. Maybe the CW Labs have been outsourced to those guys on the Internets? Hey, times are tough in tech journalism.

Anyway, color me suspicious. Suspicious enough, in fact, that I got out my stopwatch and spent half a day playing lab rat: rebuilding a shiny new Dell test system here with side-by-side XP and Vista installations.

And after I was done with more than two dozen startup/shutdown cycles for each machine, I confirmed that Vista is indeed slower than XP at startup. But I needed that stopwatch to tell the difference. On this Dell XPS410, with an Intel Core 2 Duo E6600, 4GB of RAM and a 160GB 10,000 RPM SATA drive, Windows XP Professional made it from power-on to a fully loaded desktop in an average of 61 seconds. On the same hardware, Windows Vista Business Edition needed an extra 11 seconds to get up and running. Shocking! As for shutdown times, I consistently measured nearly identical times for shutting down both operating systems. And those times are consistent with my experience on other test systems, both desktops and notebooks.

In that Computerworld article, some of the newsgroup posters complained about startup and shutdown times of five minutes or more. I can imagine that. I’ve seen similar issues with versions of Windows going back a decade or more when a bad driver or app is loaded and ties up the system at startup. In fact, I ran into the exact same problem just last month on Windows XP, caused by overaggressive security software. I never thought to blame it on the operating system, though. When that happens, it’s invariably a driver or a piece of software that’s hanging or timing out. Troubleshooting that sort of problem is tedious but relatively straightforward, and once you find the problem things get right back to normal.

But doing controlled measurements and troubleshooting doesn’t work when you have a deadline and a news hole to fill.

Editorial standards