GPU Concurrency
GPU Concurrency
复制标题
GPU并发
DOI：
10.1145/2694344.2694391
复制
发表时间：
2015
期刊：
影响因子：
--
通讯作者：
Alglave J
中科院分区：
文献类型：
--
作者：
Alglave J
并发性是普遍和令人困惑的，特别是在图形处理单元（gpu）上。目前的语言和硬件规范尚无定论；因此，程序员在编写软件时经常依赖于民间的假设。为了纠正这种状况，我们对部署的gpu的并发行为进行了大量的实证研究。有了石蕊测试（即短并发程序），我们质疑编程指南和供应商文档中关于硬件提供保证的假设。我们开发了一个工具来生成数千个石蕊测试，并在压力大的工作负载下运行它们。我们观察到一连串之前难以捉摸的弱行为，并揭露了关于GPU编程的民间信仰（通常由官方教程支持）是错误的。作为一种前进的方式，我们提出了一个Nvidia GPU硬件模型，它正确地模拟了我们在实验中看到的每一个行为。该模型是SPARC放松内存顺序（RMO）的一种变体，遵循GPU并发层次结构。
Concurrency is pervasive and perplexing, particularly on graphics processing units (GPUs). Current specifications of languages and hardware are inconclusive; thus programmers often rely on folklore assumptions when writing software.To remedy this state of affairs, we conducted a large empirical study of the concurrent behaviour of deployed GPUs. Armed with litmus tests (i.e. short concurrent programs), we questioned the assumptions in programming guides and vendor documentation about the guarantees provided by hardware. We developed a tool to generate thousands of litmus tests and run them under stressful workloads. We observed a litany of previously elusive weak behaviours, and exposed folklore beliefs about GPU programming---often supported by official tutorials---as false.As a way forward, we propose a model of Nvidia GPU hardware, which correctly models every behaviour witnessed in our experiments. The model is a variant of SPARC Relaxed Memory Order (RMO), structured following the GPU concurrency hierarchy.