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

Citations

149	Accelerating large graph algorithms on the GPU using CUDA. - Harish, Narayanan - 2007 (Show Context) Citation Context ...edominant bottleneck in this step. A similar speedup is observed in the document clustering step. 6 Related Work A myriad of work with respectable results has been reported in stand-alone GPU systems =-=[9, 12, 14, 5, 10]-=-. As more and more applications report speedups by dataparallel co-processing, the community is beginning to investigate the potential of massively data-intensive applications, such as text mining, cl...
127	OpenMP to GPGPU: a compiler framework for automatic translation and optimization,” in - Lee, Min, et al. - 2009 (Show Context) Citation Context ...PU kernels. Our programming model employs a hybrid parallelization model at the MPI process level, where parallel executions are forked and joined implicitly. This approach differs from previous work =-=[11]-=- that compiles existing OpenMP programs to CUDA kernel codes running in single node. 7 Conclusion In this paper, we presented a novel programming model for massive data parallelism with data dependenc...
105	Gpu Cluster for High Performance Computing,” - Fan, Qiu, et al. - 2004 (Show Context) Citation Context ...ieve scalability for industrial usage. In contrast, few experiments on GPU clusters can be found in literature. Two ad-hoc approaches including acceleration of scientific computations in GPU clusters =-=[7, 6]-=-. Recent work [16] proposes an MPI-like message passing framework, DCGN, for data-parallel architectures. This enables programmers to initiate message passing from inside the GPU kernel. The data sync...
77	MCUDA: An Efficient Implementation of CUDA Kernels on Multi-cores. - Stratton, Stone, et al. - 2008 (Show Context) Citation Context ...rogramming interface to GPUs, lowers the barrier for general application programmers to program on GPUs. CUDA predominantly runs on NVIDIA’s GPUs but has also been ported to generalpurpose multicores =-=[15]-=-. OpenCL, an open industry standard, tries to provide a vendor-independent programming model for all processors not restricted to GPUs. Even though it is much easier to program on GPUs than ever, cert...
45	GPUGems 3. - Nguyen, ed - 2007 (Show Context) Citation Context ...edominant bottleneck in this step. A similar speedup is observed in the document clustering step. 6 Related Work A myriad of work with respectable results has been reported in stand-alone GPU systems =-=[9, 12, 14, 5, 10]-=-. As more and more applications report speedups by dataparallel co-processing, the community is beginning to investigate the potential of massively data-intensive applications, such as text mining, cl...
28	A Flocking Based Algorithm for Document Clustering Analysis", - Cui, Gao, et al. - 2006 (Show Context) Citation Context ...ument Clustering We assess this programming model using a real-world application, namely a massive document clustering problem. The complete algorithm is based on previous research performed at ORNL (=-=[13, 4]-=-) and consists of two phases: the Tf-Icf step and flocking-based document clustering simulation. In the Tf-Icf step, the input is a corpus of plain-text documents. The goal is to reduce each document ...
28	ICF: A new term weighting scheme for clustering dynamic data streams, in: - Reed, Jiao, et al. - 1236 (Show Context) Citation Context ...ument Clustering We assess this programming model using a real-world application, namely a massive document clustering problem. The complete algorithm is based on previous research performed at ORNL (=-=[13, 4]-=-) and consists of two phases: the Tf-Icf step and flocking-based document clustering simulation. In the Tf-Icf step, the input is a corpus of plain-text documents. The goal is to reduce each document ...
27	Message passing on data-parallel architectures,” - Stuart, Owens - 2009 (Show Context) Citation Context ... industrial usage. In contrast, few experiments on GPU clusters can be found in literature. Two ad-hoc approaches including acceleration of scientific computations in GPU clusters [7, 6]. Recent work =-=[16]-=- proposes an MPI-like message passing framework, DCGN, for data-parallel architectures. This enables programmers to initiate message passing from inside the GPU kernel. The data synchronization for co...
18	Yang - Parallel Data Mining on Graphics Processors, - Fang, Lau, et al. - 2008 (Show Context) Citation Context ...ns report speedups by dataparallel co-processing, the community is beginning to investigate the potential of massively data-intensive applications, such as text mining, clustering and classifications =-=[18, 8, 17]-=-. In most cases, the on-board physical memory is one of the major constraints to achieve scalability for industrial usage. In contrast, few experiments on GPU clusters can be found in literature. Two ...
17	N-body simulations on GPUs - Elsen, Vishal, et al. - 2007 (Show Context) Citation Context ...ieve scalability for industrial usage. In contrast, few experiments on GPU clusters can be found in literature. Two ad-hoc approaches including acceleration of scientific computations in GPU clusters =-=[7, 6]-=-. Recent work [16] proposes an MPI-like message passing framework, DCGN, for data-parallel architectures. This enables programmers to initiate message passing from inside the GPU kernel. The data sync...
16	Clustering billions of data points using GPUs - Wu, Zhang, et al. - 2009 (Show Context) Citation Context ...ns report speedups by dataparallel co-processing, the community is beginning to investigate the potential of massively data-intensive applications, such as text mining, clustering and classifications =-=[18, 8, 17]-=-. In most cases, the on-board physical memory is one of the major constraints to achieve scalability for industrial usage. In contrast, few experiments on GPU clusters can be found in literature. Two ...
15	Accelerating Reed-Solomon Coding in RAID Systems with GPUs - Curry, Skjellum, et al. - 2008 (Show Context) Citation Context ...edominant bottleneck in this step. A similar speedup is observed in the document clustering step. 6 Related Work A myriad of work with respectable results has been reported in stand-alone GPU systems =-=[9, 12, 14, 5, 10]-=-. As more and more applications report speedups by dataparallel co-processing, the community is beginning to investigate the potential of massively data-intensive applications, such as text mining, cl...
6	Accelerating matlab with CUDA,” - Fatica, Jeong - 2013 (Show Context) Citation Context ...edominant bottleneck in this step. A similar speedup is observed in the document clustering step. 6 Related Work A myriad of work with respectable results has been reported in stand-alone GPU systems =-=[9, 12, 14, 5, 10]-=-. As more and more applications report speedups by dataparallel co-processing, the community is beginning to investigate the potential of massively data-intensive applications, such as text mining, cl...
5	GPU-Accelerated Text Mining - Zhang, Mueller, et al. - 2009 (Show Context) Citation Context ...ns report speedups by dataparallel co-processing, the community is beginning to investigate the potential of massively data-intensive applications, such as text mining, clustering and classifications =-=[18, 8, 17]-=-. In most cases, the on-board physical memory is one of the major constraints to achieve scalability for industrial usage. In contrast, few experiments on GPU clusters can be found in literature. Two ...
3	Geometric algorithms on cuda. - Ruiz, Ortega - 2008 (Show Context) Citation Context ...edominant bottleneck in this step. A similar speedup is observed in the document clustering step. 6 Related Work A myriad of work with respectable results has been reported in stand-alone GPU systems =-=[9, 12, 14, 5, 10]-=-. As more and more applications report speedups by dataparallel co-processing, the community is beginning to investigate the potential of massively data-intensive applications, such as text mining, cl...