Optimize for 'double'

Dear All

I am wondering is there any special optimization in MATLAB for double arrays? I have the feeling that UINT is a little slower than double array, although the reduced memory space is considerable. I need to make a trade off between speed and space now.

Thanks!

[Yingke]

For example: the range of A is 0-2^32. The operation on A is essential, + - * /.

Shall I choose Uint32 or DOUBLE?

Answer 1

[Walter Roberson]

These days it is common for CPUs to have faster floating point instructions than integer instructions.

For larger double arrays, I believe the optimized LAPACK or similar libraries may be invoked; I am not sure if those are invoked for other data types.

Answer 2

[deniel munthe]

you can do in order to normalize the data into smaller data and data processing will be faster.

[Yingke]

Thank you for your quick reply.

However, my data consist of empirical frequency, and normalization will introduce extra error.