VLCs: Managing Parallelism with Virtualized Libraries

TL;DR

This paper introduces Virtual Library Contexts (VLCs), a method to manage resource contention among libraries in parallel applications without modifying library code, leading to significant performance improvements.

Contribution

VLCs provide a novel way to control resource allocation among libraries, enabling parallelism and resource partitioning without library or OS modifications.

Findings

VLCs achieve up to 2.85x speedup on benchmarks.

VLCs enable parallel execution of thread-unsafe libraries.

VLCs effectively manage resource contention in complex parallel applications.

Abstract

As the complexity and scale of modern parallel machines continue to grow, programmers increasingly rely on composition of software libraries to encapsulate and exploit parallelism. However, many libraries are not designed with composition in mind and assume they have exclusive access to all resources. Using such libraries concurrently can result in contention and degraded performance. Prior solutions involve modifying the libraries or the OS, which is often infeasible. We propose Virtual Library Contexts (VLCs), which are process subunits that encapsulate sets of libraries and associated resource allocations. VLCs control the resource utilization of these libraries without modifying library code. This enables the user to partition resources between libraries to prevent contention, or load multiple copies of the same library to allow parallel execution of otherwise thread-unsafe code within the same process. In this paper, we describe and evaluate C++ and Python prototypes of VLCs. Experiments show VLCs enable a speedup up to 2.85x on benchmarks including applications using OpenMP, OpenBLAS, and LibTorch.