A Programming Model for Massive Data Parallelism with Data Dependencies ∗

DMCA

A Programming Model for Massive Data Parallelism with Data Dependencies ∗

Cached

Download Links

by Yongpeng Zhang , Frank Mueller , Xiaohui Cui , Thomas Potok

Citations:

1 - 1 self

BibTeX

@MISC{Zhang_aprogramming,
    author = {Yongpeng Zhang and Frank Mueller and Xiaohui Cui and Thomas Potok},
    title = {A Programming Model for Massive Data Parallelism with Data Dependencies ∗},
    year = {}
}

OpenURL

Abstract

Accelerating processors can often be more cost and energy effective for a wide range of data-parallel computing problems than general-purpose processors. For graphics processor units (GPUs), this is particularly the case when program development is aided by environments, such as NVIDIA’s Compute Unified Device Architecture (CUDA), which dramatically reduces the gap between domainspecific architectures and general-purpose programming. Nonetheless, general-purpose GPU (GPGPU) programming remains subject to several restrictions. Most significantly, the separation of host (CPU) and accelerator (GPU) address spaces requires explicit management of GPU memory resources, especially for massive data parallelism that well exceeds the memory capacity of GPUs. One solution to this problem is to transfer data between GPU and host memories frequently. In this work, we investigate another approach. We run massively data-parallel applications on GPU clusters. We further propose a programming model for massive data parallelism with data dependencies for this scenario. Experience from micro benchmarks and real-world applications shows that our model provides not only ease of programming but also significant performance gains. 1

Keyphrases

massive data parallelism programming model data dependency several restriction significant performance gain gpu memory resource data-parallel application general-purpose gpu graphic processor unit general-purpose programming explicit management gpu cluster compute unified device architecture domainspecific architecture address space general-purpose processor memory capacity program development real-world application host memory wide range micro benchmark

DMCA