MATLAB’s Racing Jacket
First Published in Automated Trader Magazine Issue 16 Q1 2010
For those conducting intensive calculations on large data sets, recent advances in GPU computing have been little short of a revelation; no more waiting days for results, or having to invest in your own giant cluster or supercomputer. On that basis, GPU computing seems a natural fit with quants’ favourite MATLAB, which probably explains why the people at Accelereyes dreamt up their Jacket application for GPU-enabling MATLAB. Andy Webb takes it for a gallop round the Automated Trader track.
Anyone opening Automated Trader for the first time wouldn't take long to realise that our readers spend quite a lot of their time thinking about speed, which means that we all spend a corresponding amount of time writing about it. Until now, much of our focus has been on speed at the point of trade execution; fastest connectivity, colocation, in-processor-cache matching engines - and so on. We haven't expended so many words on speeding up offline calculations, such as the severe number ...
