memory usage issue
in [Matlab]
|
From: ashutosh srivastava on 19 Jul 2010 12:08 dear readers, i'm in need to know that how much system-memory is being utilised in executing a particular *.m file. is there any command for the same, or is there any suitable method to know the computational complexity involved therein. ----------- ashutosh srivastava
From: Walter Roberson on 19 Jul 2010 12:32 ashutosh srivastava wrote: > i'm in need to know that how much system-memory is being utilised in > executing a particular *.m file. > is there any command for the same, > or is there any suitable method to > know the computational complexity involved therein. See my reply in http://www.mathworks.com/matlabcentral/newsreader/view_thread/286357 > or is there any suitable method to > know the computational complexity involved therein. If your memory storage is indefinite precision but each element to be stored has a finite range, then you only need a single memory location to store everything. You can even goedelize the values if using the least infinite memory is important somehow. If your memory storage is finite precision, then unless you have an infinite number of such locations you will eventually run out of memory on a large enough (non-trivial) problem. However, whether you *will* run out of memory or not for a particular run depends upon implementation details in addition to algorithmic complexity. For example, your algorithmic complexity might be completely linear in the problem size, but you might run out of actual memory if you represent the data as double precision in a situation that single precision would be adequate. About this point you should read my earlier reply.... Go ahead, I'll wait. http://i51.photobucket.com/albums/f373/wildflowerco62/EATERY/hands_twiddling_thumbs_lg_nwm.gif Conclusion: "system memory usage" is a totally inadequate method of measuring algorithmic complexity.
|
Pages: 1 Prev: QR factorization vs Least Squares Next: Fast Fourier Transformation / Power Spectral Density |