High-Level and Efficient Stream Parallelism on Multi-core Systems with SPar for Data Compression Applications

Abstract

The stream processing domain is present in several real-world applications that are running on multi-core systems. In this paper, we focus on data compression applications that are an important sub-set of this domain. Our main goal is to assess the programmability and efficiency of domain-specific language called SPar. It was specially designed for expressing stream parallelism and it promises higher-level parallelism abstractions without significant performance losses. Therefore, we parallelized Lzip and Bzip2 compressors with SPar and compared with state-of-the-art frameworks. The results revealed that SPar is able to efficiently exploit stream parallelism as well as provide suitable abstractions with less code intrusion and code refactoring.

Stream Parallelism on the LZSS Data Compression Application for Multi-Cores with GPUs

“…Our parallel algorithms are generic enough to be used along with other stream-based parallel programming frameworks such as FastFlow, TBB, and GrPPI. Also, the stream parallelism strategies developed in this work for combining CPU and single/multi-GPU could be extend to other data compression applications [22] or real-world stream processing applications [23].…”

Compressão de Dados em Clusters HPC com Flink, MPI e SPar

“…Os autores concluem que os experimentos utilizando a SPar obtiveram melhor desempenho comparado ao Flink. No artigo [Griebler et al 2017] os autores buscaram avaliar a programabilidade e eficiência da linguagem SPar. Os autores utilizaram os compressores Lzip e Bzip2, que revelaram que o SPar é capaz de explorar com eficiência o paralelismo de stream, bem como fornecer abstrações adequadas com menos intrusão de código e refatoração de código.…”

High-Level Stream Parallelism Abstractions with SPar Targeting GPUs

“…Alternatively to these options, the SPar 2 domain-specific language [9] provides a productive parallel programming model without adding significant performance overheads for multi-cores [11]. Although that SPar is demonstrating a good compromise between productivity and performance among different stream processing applications for multi-core architectures [11,10,12], automatic code generation for heterogeneous architectures composed of CPU and GPU is still not supported. Our recent investigations using SPar to annotate stream parallelism for multi-cores with manually programming data parallelism for GPUs have shown promising performance results [20].…”