Comments on: Example of the Matlab Precision Issue http://wei-jiang.com/programming/matlab/example-of-the-matlab-precision-issue Thu, 12 Jan 2012 07:52:00 +0000 hourly 1 http://wordpress.org/?v=3.3 By: Mike Croucher http://wei-jiang.com/programming/matlab/example-of-the-matlab-precision-issue/comment-page-1#comment-127 Mike Croucher Fri, 12 Mar 2010 17:45:47 +0000 http://wei-jiang.com/?p=1764#comment-127 The limitations of IEEE arithmetic apply to many packages, not just MATLAB. As you say, Maple and Mathematica can employ arbitrary precision arithmetic to solve problems that IEEE arithmetic cannot.<br><br>However, this extra precision comes at a cost and that cost is execution speed. IEEE arithmetic is implemented in hardware which means it's fast. VERY fast. Arbitrary precision arithmetic is implemented in software which makes it a LOT slower.<br><br>So, it all comes down to a trade-off. If IEEE arithmetic is good enough to solve your problem (and for many problems, it will be) then you are well advised to use it.<br><br>Cheers,<br>Mike The limitations of IEEE arithmetic apply to many packages, not just MATLAB. As you say, Maple and Mathematica can employ arbitrary precision arithmetic to solve problems that IEEE arithmetic cannot.

However, this extra precision comes at a cost and that cost is execution speed. IEEE arithmetic is implemented in hardware which means it's fast. VERY fast. Arbitrary precision arithmetic is implemented in software which makes it a LOT slower.

So, it all comes down to a trade-off. If IEEE arithmetic is good enough to solve your problem (and for many problems, it will be) then you are well advised to use it.

Cheers,
Mike

]]>