Boost.SIMD

Abstract

SIMD extensions have been a feature of choice for processor manufacturers for a couple of decades. Designed to exploit data parallelism in applications at the instruction level, these extensions still require a high level of expertise or the use of potentially fragile compiler support or vendor-specific libraries. While a large fraction of their theoretical accelerations can be obtained using such tools, exploiting such hardware becomes tedious as soon as application portability across hardware is required. In this paper, we describe BOOST.SIMD, a C++ template library that simplifies the exploitation of SIMD hardware within a standard C++ programming model. BOOST.SIMD provides a portable way to vectorize computation on Altivec, SSE or AVX while providing a generic way to extend the set of supported functions and hardwares. We introduce a C++ standard compliant interface for the users which increases expressiveness by providing a high-level abstraction to handle SIMD operations, an extension-specific optimization pass and a set of SIMD aware standard compliant algorithms which allow to reuse classical C++ abstractions for SIMD computation. We assess BOOST.SIMD performance and applicability by providing an implementation of BLAS and image processing algorithms.

Faster Base64 Encoding and Decoding Using AVX2 Instructions

“…Having to design and support functions for several different vectorized instruction sets is cumbersome. We could ask whether we can achieve similar results with higher-level C++ libraries such as the Generic SIMD Library [Wang et al 2014] or Boost.SIMD [Estérie et al 2014]. Programming languages like Swift or Rust also have generic SIMD packages that could be put to good use.…”

A Unified Dynamic Programming Framework for the Analysis of Interacting Nucleic Acid Strands: Enhanced Models, Scalability, and Speed

“…Dynamic programs are implemented using a generic programming paradigm employing expression templates and compile-time polymorphism; generic recursion equations capturing the details of the structural ensemble and free energy model are translated via template metaprogramming into a separate vectorized executable for calculating each physical quantity in Table . Single-threaded single instruction multiple data (SIMD) vectorization is implemented using the Boost.SIMD library . The convex optimization problem (eq 14) is solved in the dual form using an efficient trust region method using the Armadillo linear algebra library for matrix operations …”

Usuba

“…SIMD libraries. A flurry of C++ libraries provide a unified programming model for various SIMD architectures, such as Boot.SIMD [11], MIPP [8], UME::SIMD [14], Sierra [18] or VC [16]. These works are complementary to ours: our intent is to provide a language that (always) compiles to efficient bitsliced code while being amenable to formal verification.…”

“…Historically, bitslicing is a manual process. For instance, here is a snippet of a bitsliced implementation of the Data Encryption Standard (DES) by Kwan [17]: [11], r1^k[26], r2^k [3], r3^k [13] r11^k [17], r12^k [5], &l23, &l15, &l29, &l5);…”