Architectural specification for massively parallel computers: an experience and measurement‐based approach

Brightwell, Ron; Camp, William J.; Cole, Benjamin; DeBenedictis, Erik P.; Leland, Robert W.; Tomkins, James L.; Maccabe, A.B.

doi:10.1002/cpe.893

Cited by 13 publications

(6 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The LWFS-core should provide the infrastructure needed to provide controlled access to data distributed across multiple storage servers. 4. The LWFS-core should provide an "open architecture" for optional functionality that allows the client implementation to accept, reject, replace, or create additional functionality.…”

Section: The Lwfs-corementioning

confidence: 99%

“…Based on the specifications presented in [4], an I/O node can receive 6 GB/s from the network, but only output 400 MB/s to the RAID storage. Requests that arrive but cannot be processed are either buffered on the I/O node or rejected if the I/O node buffer is full.…”

Section: Data Movementmentioning

confidence: 99%

“…Efficient I/O is sometimes referred to as the "Achilles' heel" of massively-parallel processing (MPP) computing [4]. While part of the blame can be placed on the inability of the hardware advances for I/O systems to keep * Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy under contract DE-AC04-94AL85000.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Lightweight I/O for Scientific Applications

Oldfield

Ward

Riesen

et al. 2006

2006 IEEE International Conference on Cluster Computing

View full text Add to dashboard Cite

Today's high-end massively parallel processing (MPP) machines have thousands to tens of thousands of processors, with next-generation systems planned to have in excess of one hundred thousand processors. For systems of such scale, efficient I/O is a significant challenge that cannot be solved using traditional approaches. In particular, general purpose parallel file systems that limit applications to standard interfaces and access policies do not scale and will likely be a performance bottleneck for many scientific applications.In this paper, we investigate the use of a "lightweight" approach to I/O that requires the application or I/O-library developer to extend a core set of critical I/O functionality with the minimum set of features and services required by its target applications. We argue that this approach allows the development of I/O libraries that are both scalable and secure. We support our claims with preliminary results for a lightweight checkpoint operation on a development cluster at Sandia.

show abstract

Section: The Lwfs-corementioning

confidence: 99%

Section: Data Movementmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Lightweight I/O for Scientific Applications

Oldfield

Ward

Riesen

et al. 2006

2006 IEEE International Conference on Cluster Computing

View full text Add to dashboard Cite

show abstract

“…Thus, the only possible reason could be overheads resulting from communication between workers. These overheads are generally the result of computational costs of cache coherence; memory conflicts inherent to a shared-memory multiprocessing architecture like the INTEL Core i7 [60]; and memory conflicts between operating system services [61]. Moreover, since MATLAB looks to the operating system to open a pool of workers, it does not guarantee proper assignment of each worker to a single physical core/thread, which would result in exaggerated overheads from worker instances trying to communicate with (or waiting for) another instance on the same thread.…”

Section: Comparing Single Cpu and Spmd Executionmentioning

confidence: 99%

Parallel Simulation of Population Balance Model-Based Particulate Processes Using Multicore CPUs and GPUs

Prakash

Chaudhury

Ramachandran

2013

Modelling and Simulation in Engineering

View full text Add to dashboard Cite

Computer-aided modeling and simulation are a crucial step in developing, integrating, and optimizing unit operations and subsequently the entire processes in the chemical/pharmaceutical industry. This study details two methods of reducing the computational time to solve complex process models, namely, the population balance model which given the source terms can be very computationally intensive. Population balance models are also widely used to describe the time evolutions and distributions of many particulate processes, and its efficient and quick simulation would be very beneficial. The first method illustrates utilization of MATLAB's Parallel Computing Toolbox (PCT) and the second method makes use of another toolbox, JACKET, to speed up computations on the CPU and GPU, respectively. Results indicate significant reduction in computational time for the same accuracy using multicore CPUs. Many-core platforms such as GPUs are also promising towards computational time reduction for larger problems despite the limitations of lower clock speed and device memory. This lends credence to the use of highfidelity models (in place of reduced order models) for control and optimization of particulate processes.

show abstract

“…See [6] for a detailed description of the Red Storm, as well as a presentation of the employed design philosophy.…”

Section: Snl Red Storm Parametersmentioning

confidence: 99%

Lightweight storage and overlay networks for fault tolerance.

Oldfield¹

2010

View full text Add to dashboard Cite

The next generation of capability-class, massively parallel processing (MPP) systems is expected to have hundreds of thousands to millions of processors, In such environments, it is critical to have fault-tolerance mechanisms, including checkpoint/restart, that scale with the size of applications and the percentage of the system on which the applications execute. For application-driven, periodic checkpoint operations, the state-of-the-art does not provide a scalable solution. For example, on today's massive-scale systems that execute applications which consume most of the memory of the employed compute nodes, checkpoint operations generate I/O that consumes nearly 80% of the total I/O usage. Motivated by this observation, this project aims to improve I/O performance for application-directed checkpoints through the use of lightweight storage architectures and overlay networks. Lightweight storage provide direct access to underlying storage devices. Overlay networks provide caching and processing capabilities in the compute-node fabric. The combination has potential to signifcantly reduce I/O overhead for large-scale applications. This report describes our combined efforts to model and understand overheads for application-directed checkpoints, as well as implementation and performance analysis of a checkpoint service that uses available compute nodes as a network cache for checkpoint operations.3

show abstract

Architectural specification for massively parallel computers: an experience and measurement‐based approach

Cited by 13 publications

References 16 publications

Lightweight I/O for Scientific Applications

Lightweight I/O for Scientific Applications

Parallel Simulation of Population Balance Model-Based Particulate Processes Using Multicore CPUs and GPUs

Lightweight storage and overlay networks for fault tolerance.

Contact Info

Product

Resources

About