cuda-fixnum

cuda-fixnum copied to clipboard

Correctly handle errors from device code

2

Apparently modern compilers emit code that runs essentially as fast with and without exceptions _on the no-throw execution path_. So there are no performance reasons not to use exceptions for...

unzvfu

Specialise modexp for word-sized base

1

See the remark on p69 of [MCA, Section 2.6].

unzvfu

Implement complement representation for signed fixnums

1

See Handbook of Applied Cryptography, Section 14.2.1(ii).

unzvfu

Fixnum implementations should not expose `slot_layout` functionality

1

Algorithms that use a particular `fixnum_impl` should be expressed entirely in terms of the `fixnum_impl` interface; `slot_layout` operations _can_ be exposed via that interface, but they should be given semantically...

unzvfu

Provide RNS-based modnum implementation

1

unzvfu

Generate test cases using GMP's mpz_rrandomb function

1

From the function's documentation: > Generate a random integer with long strings of zeros and ones in the binary representation. Useful for testing functions and algorithms, since this kind of...

unzvfu

Run tests automatically with Travis

1

Set up CI with Travis (and perhaps Jenkins too?).

unzvfu

cmake and autodetect gpu

15

Change in Makefile to support up to Turing GPUs. [NVIDIA Reference](https://docs.nvidia.com/cuda/turing-compatibility-guide/index.html#building-turing-compatible-apps-using-cuda-10-0) WARNING: Creates binaries 7X larger and makes build 7X slower but supports 6X the GPUs families.

jkrauska