Saturday, March 21, 2009

A minute faster

I promised in my last post that I'd run a more detailed test and get back to you ...

My wife's Linux laptop is an IBM/Lenovo ThinkPad T43 (model 1875M2U, purchased in early 2005.) It has an Intel Centrino CPU (1.86 GHz) with 512 MB memory. She is running a generic install of Fedora 10, with all the default startup processes left running. This is how the system would be set up if any non-expert user had installed it. More importantly, it's what my Linux laptop would have looked like if I could run Linux at work.

My Windows laptop is a Dell Latitude D430 (purchased in late 2007 or early 2008) so it's about 3 years newer than my wife's laptop. It has an Intel Core2 CPU (1.20GHz) with 2 GB memory. I'm running Windows XP Professional with Service Pack 3, installed for me by our desktop support folks at work, using the standard "manager laptop" image. This is exactly the same image that every other manager at work is using. I don't know if any of the usual startup services have been disabled.

In short: the Linux laptop is about 3 years older than the Windows laptop. The Core2 CPU has a slightly different design than the Centrino, so you can't directly compare GHz speeds here.

At work, I'm a manager, so I typically only run things like Office and Firefox to do my work. Every morning, my startup routine is the same: boot my laptop, start Firefox, login to our webmail system and check for any urgent emails. I also login to our web-based calendar system to look at my meetings for the day. So it shouldn't be a surprise that I measure the time by which I can start doing work by how long it takes to boot my laptop and run Firefox.

This afternoon, I ran a side-by-side boot test, using a simple "minute:second" stopwatch to see how long it took to boot each system. The results are interesting:

Powering on both laptops simultaneously (a "cold boot") it took about the same amount of time for both to prompt me for the hard drive encryption password. Windows was ready for me a fraction of a second earlier, so I typed my username / password there first before typing my passphrase into the Linux laptop.

Windows felt like it was taking a long time to boot. I've previously discussed how you never really know if you've launched a program or opened a document in Windows, or merely selected it. While booting, Windows doesn't tell you what it's doing. You just get the "cylon" progress bar, so at least you know Windows hasn't hung. (Linux has a neat little progress bar while it's booting.)

Surprisingly, both laptops also prompted for login at about the same time (1:15 after boot.) But Windows was deceiving; after pressing ctrl-alt-del to start the login process, it took several seconds before the username / password dialog box came up. As a result, I had already typed in my username and password on the Linux laptop by the time Windows was really ready for me.

As soon as either laptop presented me with my desktop, I immediately launched Firefox. The Linux laptop brought up Firefox and presented me with a web page (2:10) much faster than Windows (3:07). So, about a minute faster.

As a point of interest, I was able login to our webmail system at work, read an email, and delete it in the extra time it took Windows to start Firefox. And that's on a laptop that's 3 years older.


  1. You should try and get some solid metric data, using the same computer (or identical ones), comparing the speeds of completing different tasks. Add it all up as a total annual value of production lost due to Windows' boot times, and application load times, present it to your managers, and presto: everyone at work is using Linux! Plenty of bloggers and tech review sites have done this kind of thing before, so you might only need to gather together their results. I know Ninja did some pretty solid measurements comparing various versions of OO.o to various versions of MS Office.

  2. Or, with the data here, you could say the following:

    Windows is not better than Linux (and seems to be slower) when Linux runs on a machine that can be purchased for 600.00 but the windows system on a 2600 machine

    = 2,000 per machine.

    I'll have a post pointing back to this and elaborating up tomorrow on my blog. (Things are kind of clogged right now).

  3. Again, YMMV. I had Gentoo and Win Server 2008 on my computer for a while, and what I found was that boot times were about the same for each. BUT: startup -> login screen was rather faster on Linux than on Windows, but login screen -> usable was faster on Windows. Starting KDE each login just took too much time.

    This meant that, for me, Windows won the boot wars, since I could turn it on, go to the bathroom, etc., come back, and log on and have it be ready earlier than Linux was.

  4. I should also say that I think Windows has slowed down a bit by now.

    Also, there could be a lot of startup crap that may be running in the background, depending on how competent your company's IT dept is. Not really services in the technical sense, but programs that start up when you log in.

  5. Why would you ever need to restart Linux?

  6. Hi.
    Very interesting blog.
    The only reason i still have windows is because i haven't gotten used to the idea of living without my ipod. Is there a way to use my ipod in ubuntu, without using wine?

  7. That all depends what kind of iPod it is. My 5th Generation 80GB works fine with Gtkpod for managing music, and Thinliquidfilm is an excellent video manager for it. It will convert any video that FFMPEG (the backend you likely use with Totem) can handle.


Note: Only a member of this blog may post a comment.