Memory Error when numpy.dot(A,B), where A is 250000x108 float matrix, B is transpose of A

Question

Any idea of do A dot B, without Memroy Error?

Answer 1

The result would be a 250000x250000 matrix, and would require about half a terabyte of RAM.

Did you mean numpy.dot(B, A)?

Answer 2

The dot product you are trying to calculate produces a 250000x250000 array, which will require 500Gb of memory if you are using the default double precision floating point. Are you sure that is what you need?