The Art of C++ / Tuple is a C++11 header-only proof-of-concept implementation of
std::tuple.
Why another implementation of std::tuple? To provide a proof-of-concept that,
when avoiding recursion, code can benefit in significant ways. I prepared a talk
about it, including some benchmarks.
TL;DR: GCC 5 on Linux with libstdc++'s std::tuple requires 19.6s and an instantiation
depth of at least 3.719 to compile an
example
generating a tuple with 265 elements via std::tuple_cat.
tao::tuple requires 1.2s and an instantiation depth of 26 on the same system.
Apple LLVM 7.0 (~Clang 3.7) with libc++'s std::tuple requires 70s and an instantiation
depth of 514 to compile the example. tao::tuple requires 1.7s and an instantiation depth
of 15 on the same system. This case does not change std::integer_sequence or
std::make_integer_sequence, all improvements are coming from all the other improvements
from avoiding recursion.
tao::tuple supports CUDA.
- Requires C++11 or newer.
- Tested with GCC 4.8+, Clang 3.4+, and Visual Studio 2017.
You can download and install The Art of C++ / Tuple using the Conan package manager:
conan install -r conancenter --requires="taocpp-tuple/1.0.0"
The taocpp-tuple package in conan is kept up to date by Conan team members and community contributors. If the version is out-of-date, please create an issue or pull request on the Conan Center Index repository.
Released 2018-08-01
- Initial release.
The Art of C++ is certified Open Source software. It may be used for any purpose, including commercial purposes, at absolutely no cost. It is distributed under the terms of the MIT license reproduced here.
See the LICENSE file for details.