Benchmarks Of The Gentoo-Based Calculate Linux Desktop

Written by Michael Larabel in Operating Systems on 2 August 2010 at 02:00 AM EDT. Page 2 of 6. 81 Comments.

One of the first complaints we had about Calculate Linux Desktop 10.4 was its use of an older Mesa release (v7.5.2) by default. With the rate that the various open-source graphics drivers have been maturing along with core Mesa and the Gallium3D driver architecture, Mesa 7.5.2 is quite out of date even though it was released in September of 2009. Calculate Linux Desktop 10.4 was released this past April and it certainly would have been nicer to see Mesa 7.7 or even Mesa 7.8 used as the default. Of course, users can upgrade the Mesa stack themselves on Calculate Linux Desktop from Gentoo Portage if they are not using the proprietary NVIDIA or ATI drivers. With this older Mesa stack by default, we were only able to test the Samsung NC10 with its Intel graphics but not the AMD system using the newer ATI R700 graphics with these first OpenGL benchmarks.

Calculate Linux Desktop 10.4 with Mesa 7.5.2 ran at around the same speed of Ubuntu 10.04 LTS, which used Mesa 7.7.1, but it was not quite as fast as Sabayon 5.3 that uses Mesa 7.8.1. Fedora 13 also used Mesa 7.8.1 but its frame-rate was barely better than that of Calculate / Ubuntu.

With World of Padman, Calculate Linux Desktop 10.4 and its older Mesa was actually much faster than the other distributions with this ioquake3-based game. Oddly, Sabayon and Fedora with its most up-to-date Intel DRI driver were actually much slower than Calculate and Ubuntu that is still one major release series behind. It looks like there may be some regressions entered into Mesa that are hampering the OpenGL performance in this game.

With our first non-graphics test, on the Intel Atom netbook the fastest 32-bit distribution for the 7-Zip test was Calculate Linux Desktop, but the difference between it and the three other Linux desktops were statistically insignificant. Fedora 13 was the fastest on the 64-bit AMD system, but again the results between the four tested distributions were too close to call a decisive winner.


Related Articles