Scientific Computing

MATLAB - N72 Script (http://www.mathworks.com)

MATLAB is your multipurpose scientific computing application. Every engineer and his brother has used Matlab at one point or another. It's a very flexible application used in high school to teach basic Newtonian physics and was used in industry to design the Joint Strike Fighter. This is also a single threaded application. Why? Mathworks has done their own studies and determined that for most Matlab tasks, a lot of computation is spent parsing and processing the script, something that isn't parallel at all. Parsing scripts isn't a very glamorous aspect of scientific computing, but it's very important to real-world use. Think of the car that does 0-60 in 4 seconds but requires you to refill the gas tank every 10 miles. There's no doubt that the car is fast, but no one would really use it.

To get some measure of real-world performance, I've gone with a lengthy script from our lab, which I've just left codenamed as N72. Due to proprietary technologies, I'll just leave the description brief: data from an MRI of the heart is read into memory and we process out the 3D geometry of the heart, and then "do some magic math" to figure out the position of muscle cells in the wall. This is fairly exotic stuff and the whole thing requires the system to have a gig of system RAM. This script represents a real-world example of reading in raw data and then processing it to get the meaningful data. The thing with scientific computing is that different fields of science have different tasks and different tasks will perform differently on different architectures.

It's obviously not designed to be a comprehensive Matlab performance benchmark, but it's a real-world test that has a realistic balance of true computation and script parsing performance. This is something that we feel is missing from ScienceMark. ScienceMark is better at representing the back-end calculation rather than the actual performance measures that affect user-input and response.

This is as much as I can show you

This is only one slice of what is scientific computing and each task is going to have a different performance signature for every application. That said, we would much rather develop a set of benchmarks on commercial scientific computing applications that are more likely to represent how CPUs are used instead of home-brew software. (i.e. it's the same question about buying a graphics card to run 3DMark versus buying a graphics card to play games).

Our version is Matlab 6.5 Release 13, and all scripts are in pcode.