All geek questions in one place

Arithmetic parallel library with arbitrary precision

I am trying to do modular exponentiation of integers with a very large modulus using repeated squaring (in my case, the power is always 2, so I believe this is the most efficient way). Thanks to the good property of my module, the computational remainder is cheap; the hard part is multiplication.

I am currently running GMP on the Intel Core 2 Quad. I would like to effectively use the four processor cores, but GMP does not scale in SMP environments, so I'm looking for an alternative arithmetic library of arbitrary precision. I found several libraries for parallel computing on matrices, but I really need a library for integers.

Is there what I'm looking for?

+6
source share
2 answers

Answer: yes, multithreaded libraries of arbitrary precision exist . But I do not know a single one that is actually public. (comparable speed to GMP)

For example, libraries of arbitrary accuracy, which are used in Pi-computing programs, TachusPi and y-cruncher are able to perform multi-threaded arithmetic on large numbers.

However, both libraries are closed and inaccessible to the public.

Affiliation: I am the author of y-cruncher . So I wrote one of these multi-threaded libraries of arbitrary precision.

+2
source

Have you checked http://mpir.org ? They claim to do this with the GMP option and use GPUs.

+1
source

Source: https://habr.com/ru/post/900119/

More articles:


All Articles