[Numpy-discussion] speed of array vs matrix
Mon Oct 25 09:04:48 CDT 2010
On Mon, Oct 25, 2010 at 6:48 AM, Citi, Luca <firstname.lastname@example.org> wrote:
> I have noticed a significant speed difference between the array and the matrix implementation of the dot product, especially for not-so-big matrices.
> For example:
> In : import numpy as np
> In : b = np.random.rand(104,1)
> In : bm = np.mat(b)
> In : a = np.random.rand(8, 104)
> In : am = np.mat(a)
> In : %timeit np.dot(a, b)
> 1000000 loops, best of 3: 1.74 us per loop
> In : %timeit am * bm
> 100000 loops, best of 3: 6.38 us per loop
> The results for two different PCs (PC1 with windows/EPD6.2-2 and PC2 with ubuntu/numpy-1.3.0) and two different sizes are below:
> array matrix
> 8x104 * 104x1
> PC1 1.74us 6.38us
> PC2 1.23us 5.85us
> 8x10 * 10x5
> PC1 2.38us 7.55us
> PC2 1.56us 6.01us
> For bigger matrices the timings seem to asymptotically approach.
> Is it something worth trying to fix or should I just accept this as a fact and, when working with small matrices, stick to array?
I think the fixed overhead comes from the subclassing of arrays. The
subclassing is done in Python and if an operation creates a matrix
then __array_finalize__ is called. All that adds up to overhead.
I wrote a mean-variance optimizer with matrices. Switching to arrays
gave me a big speed up.
More information about the NumPy-Discussion