is there any cuda library for doing arbitrary-precision arithmetic?

Accelerated Computing GPU-Accelerated Libraries
1

Hi,
Is there any cuda library for doing arbitrary-precision arithmetic, like for example
multiplication of 2 million-digit numbers?

Thanks much,
mani

2

There is an old thread on StackOverflow that may come useful if you are trying to roll your own: 128 bit integer on cuda? - Stack Overflow .

3

This may be of interest:

[url]https://github.com/NVlabs/xmp[/url]

I don’t happen to know if it handles million-digit numbers. There is a corresponding paper:

[url]http://www.acsel-lab.com/arithmetic/arith23/data/1616a047.pdf[/url]

One of the areas of focus is on efficient use of the XMAD instruction (e.g. in Maxwell). Newer architectures (volta and beyond) may have more performant integer instruction paths (compared to XMAD) so the library may not be as well optimized for these newer architectures. But that is based on a casual review, not a careful analysis.