Face off part two: Windows vs Linux real world RAM and disk tests

David M Williams
Sunday, 27 July 2008 19:44

Opinion and Analysis

Last week I put Windows Vista Ultimate and Fedora Linux 9 to a test. The article hit the front page of Digg but received a lot of criticism by those disappointed with the performance of Internet Explorer. So, let's dig deeper and use Firefox to see if Internet Explorer's memory footprint is actually a Windows Vista "feature" or not. It's time for the ultimate smackdown: Internet Explorer vs Firefox on Vista.

To recap, I wanted to know how these two different systems performed internally when performing functionally identical tasks using the applications generally considered as the standard on those platforms. The results spoke extremely favourably for the way Linux handles memory.

I used two laptops, both with 15” screens, 5400 RPM hard disks, 4GB RAM and dual core processors. The system running Windows Vista was gruntier due to a slightly better CPU and a dedicated video card. However, both were equally matched in terms of memory and, being 32-bit processors, 4GB was the maximum they could handle anyway.

On the Vista side, I used Internet Explorer and Microsoft Word and on the Linux side Firefox and OpenOffice Writer. Under Windows Vista the CPU usage continually fluctuates wildly even when idle. Yet, under Linux I had a consistent 1% CPU load whenever idle. Additionally, Linux only began using swap space on disk when I loaded up enough applications and data to fill the available RAM. By contrast Windows Vista began swapping out to disk even when 2GB of RAM was still available.

I also pointed to an extremely interesting test on Tom’s Hardware. In this test Windows Vista had its pagefile disabled and then the RAM was gradually decreased down to 512MB RAM. With that amount of memory Vista was totally unusable and would crash even when idle. I investigated how Fedora would operate under the same conditions, namely without a swap partition and with diminishing quantities of RAM. At no time did the system become unusable, and in fact it remained highly responsive.

To my pleasant surprise, this simple story of a simple man’s simple experiments hit the front page of Digg and, at the time of writing now, had 752 diggs. This means that 752 more people found the article of interest than those who did not. However, the story did receive some criticism and the complaints were largely reasonable.

Some pointed out the tests performed violated pure scientific methodology by not comparing like for like: they suggested that firstly, the laptops should have been equal – and that’s quite fair, I certainly wish I had two high performance laptops like the ASUS Lamborghini – but I still think it is telling that even with lesser hardware specifications Linux fared better. Secondly, however, it was stated the applications should have been equal also: that is, Firefox vs Firefox, OpenOffice vs OpenOffice. I believe I was upfront in the piece I was comparing the typically established default programs for that platform but the complaint definitely raises an interesting question. Was it truly Vista that was holding on to memory – by clever design to improve performance – or was it Internet Explorer carelessly leaking memory like a sieve?

All good questions deserve to be answered. So, here’s what happens if you compare Firefox on Windows Vista with Internet Explorer and what the results mean.

CONTINUED