Uncategorized —

Microsoft study finds Linux to have no advantage on older hardware

Microsoft has published a new study that attempts to refute the claims that …

Proponents of the open-sourced GNU/Linux operating system are quick to compare it favorably to Microsoft Windows, citing a series of advantages for their favorite OS. The first one to come up is invariably price, an area where Linux can always claim victory, as US$0 is a difficult target for any closed-source software company to beat. However, price alone is rarely a convincing argument for switching operating systems, so other benefits are quickly put forward.

One of the most pervasive claims is that Linux runs better than Windows on older or low-end hardware. Typically this argument is accompanied by a claim that Microsoft software suffers from "bloat," which increases the base hardware requirements of the operating system. This argument is often used to support deployments of the GNU/Linux operating system in settings where PC hardware has not been upgraded in many years, such as schools, public access centers and computer labs in poor countries.

But does Linux really fare better in such environments? Microsoft, eager to win any battle for the desktop, no matter how small, sponsored an internal study to determine if there was any truth to these claims. While some may be quick to dismiss the claims of any Microsoft study on the usefulness of Linux, the results were still somewhat interesting.

The study tested four hardware configurations: a "1995" box with a 486 and 16 megs of RAM, a "1997" box with a Pentium 233 and 32 megs, a "1999" box with a PII 450 and 64 megs, and a "2001" box with a PIII 800 and 128 megs. Eight current distributions of Linux were tested, including Mandrake 10, SuSE Pro 9.2, Slackware 10.1 and Knoppix 3.7. These were matched up against Windows XP (curiously misspelled as "Windows XT" in the graph) and Windows Server 2003.

None of the operating systems installed and ran successfully on the 1995 hardware. Only Slackware and Knoppix worked on the 1997 hardware. When the clock was moved ahead to 1999, all operating systems except for Linspire, Xandros, and Windows Server 2003 were able to run. Every OS ran on the 2001 hardware, although Fedora Core 3 had a driver problem with the X Server.

Curiously, if you look at these results closely, they seem to confirm the idea that Linux will run on older hardware, at least if you are talking about Slackware and Knoppix specifically. However, overall the two operating systems ended up about the same. While this does tend to discredit the idea that "Linux runs faster on older hardware," at the very least, it runs no worse.

The other point that the study brings up is that some distributions fared significantly better than others. This leads into the third major benefit that Linux fans like to tout, the diversity and customization available with Linux that is not available with Windows. If Windows XP does not run on available hardware, you are pretty much out of luck. However, with Linux, different distributions are available that might work. In an extreme case, the operating system can be individually tailored for a specific installation. Microsoft's report argues that most people do not have the expertise to customize Linux in this way, and this is a valid point, but the ability is still there.

On the flip side, many Linux distributions had difficulty with getting X Server to work with older display adapters. In my personal experience, I have found that while really old versions of Linux would work fine on my Pentium 100 with 64 megs of RAM, a 2 meg SVGA card and an ancient 14" monitor, more modern ones simply refuse to configure X for graphical display and no amount of fiddling with configuration files can fix it. The problem of getting display drivers for old hardware is a valid concern, and Windows actually does fairly well in this respect.

However, in the end, it may all be a moot point. With new computers reaching the sub-US$300 level, most 1997-era hardware finds itself in a landfill, rather than in productive use. While it might be nice to tout the ability of Linux (certain distributions, at least) of running on older hardware, in practice this is something that is rarely done. And both platforms seem equally at fault in the "bloatware" category. I recently saw a magazine that came with a distribution of SuSE Linux (bundled with OpenOffice.org and many other free applications) being offered on a dual-sided, 8 gigabyte DVD. Contrast this with my AmigaOne, where the OS takes up only 30 megabytes in total and a whole slew of applications can fit on one 512 megabyte flash card.

Channel Ars Technica