topic Any long integer packages ? in Intel® oneAPI Math Kernel Library

Any long integer packages ?

WSinc — Sat, 26 Jan 2019 01:37:33 GMT

I need to work with REALLY LARGE integers.

What would be ideal is something that allows any length of integer.

The basic operations on those, of course. + - / and *

Has anyone developed those ?

MKL suggests 32/64 bit

Gennady_F_Intel — Sat, 26 Jan 2019 05:33:21 GMT

MKL suggests 32/64 bit integer only. You may try to take the GMP library for arbitrary precision arithmetic.

arbitrary precision is NOT

WSinc — Sat, 26 Jan 2019 09:01:42 GMT

arbitrary precision is NOT THE SAME as long integer.

Bill, you would have better

mecej4 — Sun, 27 Jan 2019 11:34:00 GMT

Bill, you would have better luck with your question in a more appropriate forum, such as the Intel Fortran forums, or in Usenet comp.lang.fortran.

Secondly, you would need to give a more detailed description of what your needs are, since phrases such as "really large" have very little meaning.

The following page contains a number of links that you may examine to see if the available software there would work for you: http://crd-legacy.lbl.gov/~dhbailey/mpdist/ (considering integers as a subset of reals).

See also https://www.fortran.com/big_integer_module.f95 m for an integers-only solution. You will need to know the maximum integer beforehand with this module.

I thought LONG INTEGER would

WSinc — Wed, 13 Feb 2019 19:47:00 GMT

I thought LONG INTEGER would be self-explanatory.

I was referring to integers longer than the default length of integer(8)

Now one could get 15 byte integers by declaring variables as real(16).

but that is a sloppy and messy way of going about it.

These are used in cryptography and other branches of mathematics BTW.