stochasticflux

Linux vs. OS X on a Macbook

About a year ago I decided to finally pick up a laptop to more easily facilitate bringing my work home. After looking around at the options that were available I ended up with a Macbook. It simply appeared to be the cheapest solution given the specifications I wanted in a 13″ notebook. For the first year I worked with OS X as I didn’t have the time to fiddle with Ubuntu to get everything working properly. Though now with the upcoming release of Gusty Gibbon I decided to give the improved laptop support a shot. I and I have to say I am really excited. Nearly everything worked right out of the box, it was even easy to setup my apple bluetooth mouse and keyboard. My only grips are the overly sensitive touchpad (this supposedly can be fixed through adjusting the settings) and the lack of automatic recognition of external monitor’s resolution.

Though I digress, the point of this post is that I have seen a number of discussions that Linux on the Macbook appears to be more responsive. The Macbook is a great platform and is extremely fast when completing regular tasks under OS X or Ubuntu so I didn’t want to just speculate on performance. Instead, I wanted to actually measure the difference in performance. For me, the vast majority of my processor intensive computing is done via MATLAB and therefore this is where I looked for a performance comparison.

For reference I have a 2.0 Ghz Intel Core Duo Macbook with 2 GB of RAM. I started up each operating system and right off of the back ran MATLAB 7.4 R2007a and the following benchmark code:

x = bench(10); mean(x)

Output under OS X:

ans =

0.3091 0.4460 0.3285 1.4133 1.0258 0.4507

Output under Ubuntu 7.10

ans =

0.3037 0.4127 0.2488 0.5628 0.4634 0.4018

As may be clearly seen, MATLAB under Ubuntu on the Macbook outperformed the MATLAB instillation on OS X. I have yet to fully test out the performance when running some of my current research but these benchmarks provide some hope that I may spend less time waiting.

Add comment October 11th, 2007

octave + aqua terminal + gnuplot on OS X

Installing Octave on OS X (intel) with gnuplot.

1.) Install the pre-built binaries for octave from HPC Mac OS X

2.) Install aqua term

3.) The pre-built binaries for gnuplot do not seem to work adequatley with aqua term and octave at the current time, so download the source for the newest version of gnuplot

4.) Untar the source files and open a terminal shell in the new directory and excute the following commands*

export CFLAGS='-arch i386'
export LDFLAGS='-arch i386'
./configure --with-readline=builtin
make
sudo make install

Afterwords, every thing should be working fine, test it out in octave with:

plot((-10:10).^2)

* Thanks to Nick Vox for this information.

Add comment April 12th, 2007