libsimdpp is a portable header-only zero-overhead C++ wrapper around single-instruction multiple-data (SIMD) intrinsics found in many compilers. The library presents a single interface over several instruction sets. In addition to that, the library allows to compile the same source code for different instruction sets and seamlessly hook the implementations to an internal dynamic dispatch mechanism.
The library resolves differences between instruction sets by implementing the missing functionality as a combination of several intrinsics. Moreover, the library supplies a lot of additional, commonly used functionality, such as various variants of matrix transpositions, interleaving loads/stores, optimized compile-time shuffling instructions, etc. Each of these are implemented in the most efficient manner for the target instruction set.
The library sits somewhere in the middle between programming directly in intrinsics and even higher-level SIMD libraries. As much control as possible is given to the developer, so that it's possible to exactly predict what code the compiler will generate.
The library supports SSE2, SSE3, SSSE3, SSE4.1, AVX, AVX2, FMA3, FMA4, XOP and NEON instruction sets.
Code for ALTIVEC instruction set has been added, but it has not been tested on actual hardware, thus definitely contains bugs.
The primary development of the library happens in C++11. A C++98-compatible version of the library is provided on the cxx98 branch. Boost libraries are used as a replacement for C++11-only functionality. The library does not depend on boost if the built-in dispatcher functionality is not used. The library has been tested with the following compilers:
-
C++11 version:
- GCC: 4.8.1, 4.7.3
- Clang: 3.3, 3.4
-
C++98 version
- GCC: 4.8.1, 4.7.3
- Clang: 3.3, 3.4
The C++98 version has been tested with boost 1.53 and 1.54. Most probably other versions will work too.
Online documentation is provided here.
Due to [bug #28831] (http://gcc.gnu.org/bugzilla/show_bug.cgi?id=28831), GCC produces non-optimal code for double precision floating point math on ARM. Clang does not have this problem.
- Implement an interface that would allow the library to seamlessly choose the vector type that is most optimal for the target instruction set. Currently the vector width is selected explicitly, which doesn't produce optimal code when the same code is compiled to an architecture with wider or narrower vector unit.
The library is distributed under two-clause BSD license:
Copyright (C) 2011-2013 Povilas Kanapickas tir5c3@yahoo.co.uk All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.