Optimise the crossprod routine for a single dge matrix


This computes the optimal crossover point between the PLASMA (CPU) and MAGMA (GPU) libraries. For smaller matrices PLASMA is generally used and for larger matrices MAGMA will be used. This function calculates this for crossprod function in this package and saves it for future use.


OptimisecrossprodDge(increment = 128, verbose = FALSE)



This is the step between the problem sizes being calculate. We compare the timings at different problem sizes for the PLASMA and MAGMA implementations. This is the size between those problem sizes. A smaller increment will give a more accurate crossover point, however, 128 should be sufficient.


This displays the timing information for the operations if it is set to TRUE. It is FALSE by default.


This simply runs the crossprod function, for a dge matrix, testing it numerous times at different problem sizes. It benchmarks the PLASMA and MAGMA libraries against each other. For smaller problem sizes PLASMA will be faster, as the size increases the MAGMA library will become more optimal. This routine will find and save that optimal point for future use.



Questions? Problems? Suggestions? or email at ian@mutexlabs.com.

Please suggest features or report bugs with the GitHub issue tracker.

All documentation is copyright its authors; we didn't write any of that.