Engineering Papers⌕ Search

Engineering topics

Marquez, Andres

Publications and source records attributed to Marquez, Andres.

51 records · Page 3

Designing Energy Efficient Communication Runtime Systems: A View from PGAS Models

As the march to the Exascale computing gains momentum, energy consumption of supercomputers has emerged to be the critical roadblock. While architectural innovations are imperative in achieving computing of this scale, it is largely dependent on the systems software to leverage the architectural innovations. Parallel applications in many computationally intensive domains have been designed to leverage these supercomputers, with legacy two-sided communication semantics using Message Passing Interface. At the same time, Partitioned Global Address Space Models are being designed which provide global address space abstractions and one-sided communication for exploiting data locality and communication optimizations. PGAS models rely on one-sided communication runtime systems for leveraging high-speed networks to achieve best possible performance. In this paper, we present a design for Power Aware One-Sided Communication Library - PASCoL. The proposed design detects communication slack, leverages Dynamic Voltage and Frequency Scaling (DVFS) and Interrupt driven execution to exploit the detected slack for energy efficiency. We implement our design and evaluate it using synthetic benchmarks for one-sided communication primitives, Put, Get and Accumulate and uniformly non-contiguous data transfers. Our performance evaluation indicates that we can achieve significant reduction in energy consumption without performance loss on multiple one-sided communication primitives. The achieved results are close to the theoretical peak available with the experimental test bed.

Vishnu, Abhinav↗

Evaluating In-Clique and Topological Parallelism Strategies for Junction Tree-Based Bayesian Inference Algorithm on the Cray XMT

Long viewed as a strong statistical inference technique, Bayesian networks have emerged to be an important class of applications for high-performance computing. We have applied an architecture-conscious approach to parallelizing the Lauritzen-Spiegelhalter Junction Tree algorithm for exact inferencing in Bayesian networks. In optimizing the Junction Tree algorithm, we have implemented both in-clique and topological parallelism strategies to best leverage the fine-grained synchronization and massive-scale multithreading of the Cray XMT architecture. Two topological techniques were developed to parallelize the evidence propagation process through the Bayesian network. One technique involves performing intelligent scheduling of junction tree nodes based on its topology and relative size. The second technique involves decomposing the junction tree into a much finer tree-like representation to offer much more opportunities for parallelism. We evaluate these optimizations on five different Bayesian networks and report our findings and observations. Another important contribution of this paper is to demonstrate the application of massive-scale multithreading for load balancing and use of implicit parallelism-based compiler optimizations in designing scalable inferencing algorithms.

99 GENERAL AND MISCELLANEOUS↗

Fault Oblivious eXascale Whitepaper

In this paper we present a software system which supports dynamic, irregular, adaptive applications. Data objects are created and structured in a hierarchical manner, with replication as needed to provide a high degree of redundancy. The data objects can contain data, code, tasks (work descriptors with references to data, code, and other tasks) and higher level structures such as work queues. The higher level structures benefit from the properties of the data objects: redundant storage to support resiliency in the face of hardware failure; hierarchical structure to optimize use of the HPC system; and a presence of object names, available in the per-user file system name space, which allows any application, not just specially written HPC applications, to make use of the data even while it is on the HPC system. Our use of hierarchy will make the runtime scalable to very large systems. Our use of redundancy will allow programs to be written in a fault-oblivious manner, eliminating the need for system-level checkpointing. Putting data object names into the file system name space allows for interactive use of the system by users. With this approach, we will be able to finally leave the batch era behind, a half-century after the invention of time sharing. We will be able to stop bounding program through- put by the checkpoint interval. Application data will be accessible at any time, not hidden behind opaque 128-bit pointers or MPI ranks, but given a name that is visible everywhere. Programmers can stop laying out data, and thinking about where the data is, and the code is, and the nodes are, and stick with the problem of what the application is supposed to be doing. This work, if it succeeds, will enable scientific computing to scale to the next generation of machines.

99 GENERAL AND MISCELLANEOUS↗

Designing Energy Efficient Communication Runtime Systems for Data Centric Programming Models

The insatiable demand of high performance computing is being driven by the most computationally intensive applications such as computational chemistry, climate modeling, nuclear physics, etc. The last couple of decades have observed a tremendous rise in supercomputers with architectures ranging from traditional clusters to system-on-a-chip in order to achieve the petaflop computing barrier. However, with advent of petaflop-plus computing, we have ushered in an era where power efficient system software stack is imperative for execution on exascale systems and beyond. At the same time, computationally intensive applications are exploring programming models beyond traditional message passing, as a combination of Partitioned Global Address Space (PGAS) languages and libraries, providing one-sided communication paradigm with put, get and accumulate primitives. To support the PGAS models, it is critical to design power efficient and high performance one-sided communication runtime system. In this paper, we design and implement PASCoL, a high performance power aware one-sided communication library using Aggregate Remote Memory Copy Interface (ARMCI), the communication runtime system of Global Arrays. For various communication primitives provided by ARMCI, we study the impact of Dynamic Voltage/Frequency Scaling (DVFS) and a combination of interrupt (blocking)/polling based mechanisms provided by most modern interconnects. We implement our design and evaluate it with synthetic benchmarks using an InfiniBand cluster. Our results indicate that PASCoL can achieve significant reduction in energy consumed per byte transfer without additional penalty for various one-sided communication primitives and various message sizes and data transfer patterns.

99 GENERAL AND MISCELLANEOUS↗

Fault Detection in Distributed Climate Sensor Networks using Dynamic Bayesian Networks

The Atmospheric Radiation Measurement program operated by U.S. Department of Energy is one of the largest climate research programs dedicated to the collection of long-term continuous measurements of cloud properties and other key components of the earth’s climate system. Given the critical role that collected ARM data plays in the analysis of atmospheric processes and conditions and in the enhancement and evaluation of global climate models, the production and distribution of high-quality data is one of ARM’s primary mission objectives. Fault detection in ARM’s distributed sensor network is one critical ingredient towards maintaining high quality and useful data. We are modeling ARM’s distributed sensor network as a dynamic Bayesian network where key measurements are mapped to Bayesian network variables. We then define the conditional dependencies between variables by discovering highly correlated variable pairs from historical data. The resultant dynamic Bayesian network provides an automated approach to identifying whether certain sensors are malfunctioning or failing in the distributed sensor network. A potential fault or failure is detected when an observed measurement is not consistent with its expected measurement and the observed measurements of other related sensors in the Bayesian network. We present some of our experiences and promising results with the fault detection dynamic Bayesian network.

54 ENVIRONMENTAL SCIENCES↗

Applications in Data-Intensive Computing

This book chapter, to be published in Advances in Computers, Volume 78, in 2010 describes applications of data intensive computing (DIC). This is an invited chapter resulting from a previous publication on DIC. This work summarizes efforts coming out of the PNNL's Data Intensive Computing Initiative. Advances in technology have empowered individuals with the ability to generate digital content with mouse clicks and voice commands. Digital pictures, emails, text messages, home videos, audio, and webpages are common examples of digital content that are generated on a regular basis. Data intensive computing facilitates human understanding of complex problems. Data-intensive applications provide timely and meaningful analytical results in response to exponentially growing data complexity and associated analysis requirements through the development of new classes of software, algorithms, and hardware.

99 GENERAL AND MISCELLANEOUS↗

Energy Efficient and Sustainable HPC at Pacific Northwest National Laboratory

Power consumption in US data centers has been escalating at an alarming rate. In response to Public Law 109-431, EPA reported that electricity usage by US data centers accounted for 1.5% of the total electricity used in the US in 2006. They projected that if current data center operating practices continue, electricity usage will almost double to 2.9% of the total electricity used in the US in 2011 (EPA, 2007). Consequently EPA issued a call to action to both government and industry to collaborate and set aggressive goals to reduce power consumption in data centers. In the spirit of responding to the findings and recommendations of the EPA report, the Pacific Northwest National Laboratory (PNNL) has teamed with several key organizations including: The Green Grid (TGG), ASHRAE TC9.9, IBM, 3M, and SprayCool. As part of this effort, a highly instrumented liquid-cooled cluster has been installed at PNNL. The cluster is housed in an 800 ft2 data center which resides in a mixed-use data center with a significant amount of instrumentation as well. The eventual objective of the effort is to be able to report the real-time power consumption, energy efficiency, and productivity of the liquid-cooled data center. Preliminary results from the effort at PNNL are reported in this paper. Thermal results are reported for the hottest server components, including the microprocessors and memory DIMMs. Under all conditions tested, the components have not exceeded manufacturers’ specifications. More importantly, the data show that the liquid-cooled servers can be maintained within specifications while rejecting to non-chilled facility water at 78°F (25.6°C). Furthermore, a reasonable extrapolation suggests that the specifications can still be maintained at 86°F (30°C). In an effort to address global warming, work has started on the qualification of a new 3M Fluoroketone fluid that has a Global Warming Potential (GWP) of 1. This GWP is the lowest published value of all commercially available coolants. Details are provided in the body of this paper.

54 ENVIRONMENTAL SCIENCES↗

Implementing and Evaluating Multithreaded Triad Census Algorithms on the Cray XMT

Commonly represented as directed graphs, social networks depict relationships and behaviors among social entities such as people, groups, and organizations. Social network analysis denotes a class of mathematical and statistical methods designed to study and measure social networks. Beyond sociolo-gy, social network analysis methods are being applied to other types of data in other domains such as bioinformatics, computer networks, national security, and economics. For particular problems, the size of a social network can grow to millions of nodes and tens of millions of edges or more. In such cases, researchers could benefit from the application of social network analysis algorithms on high-performance architectures and systems. The Cray XMT is a third generation multithreaded system based on the Cray XT-3/4 platform. Like most other multithreaded architectures, the Cray XMT is designed to tolerate memory access latencies by switching context between threads. The processors maintain multiple threads of execution and util-ize hardware-based context switching to overlap the memory latency incurred by any thread with the computations from other threads. Due to its memory latency tolerance, the Cray XMT has the poten-tial of significantly improving the execution speed of irregular data-intensive applications such as those found in social network analysis. In this paper, we describe our experiences in developing and optimizing three implementations of a social network analysis method known as triadic analysis to execute on the Cray XMT. The three im-plementations possess different execution complexities, qualities, and characteristics. We evaluate how the various attributes of the codes affect their performance on the Cray XMT. We also explore the effects of different compiler options and execution strategies on the different triadic analysis im-plementations and identify general XMT programming issues and lessons learned.

45 MILITARY TECHNOLOGY, WEAPONRY, AND NATIONAL DEF↗

Real-time Data Center Energy Efficiency At Pacific Northwest National Laboratory

The escalating consumption of power in data centers worldwide has brought the issue of data center energy efficiency to the forefront. Data center owners and operators now regard detailed knowledge of the energy efficiencies of their data centers as a competitive advantage. With funding from the Department of Energy (NNSA), PNNL has undertaken an in-depth analysis of the real-time energy efficiency for its Energy Smart Data Center Test Bed(ESDC-TB), which is housed in the mixed-use EMSL. The analysis is centered around the real-time display of The Green Grid’s proposed DCiE metric. To calculate this metric, PNNL relies on a variety of sources of data. At the ESDC-TB level, the data center is instrumented to the 100% level (all power consumption, and water temperatures and flow rates are measured). Most of this data is monitored in real-time, but the exception to this is with the CRAHs, which rely on a one-time power consumption measurement for the blowers (these are single speed blowers, so a one-time measurement suffices.). Outside of the data center (EMSL facility level), PNNL relies on the following: • Real-time data from the entire chiller plant (five chillers), six chilled water pumps, and one of four cooling towers (blowers only). • One-time power measurements for a single fixed speed pump that is representative of each grouping of pumps (the other pumps are assumed to possess the same power consumption levels). • One-time power measurements for a single two-speed cooling tower blower. This same blower model is deployed in three of the four cooling towers, so is assumed to be representative for all these blowers. • One-time power measurements for a single fixed speed cooling tower pump. This same pump model is deployed in all four cooling towers, so is assumed to be representative for all these pumps. A software tool named FRED was developed by PNNL to acquire, reduce, display, and archive all the data acquired from the ESDC-TB and EMSL. FRED provides the ability to display various levels of real-time data starting at the ESDC-TB and EMSL levels, then to lower levels as desired. For example, for the ESDC-TB, graphical screens are provided at the data center level, the rack level, the server level, and even the component level. In the near-term, FRED will also display the real-time DCiE. One of the major challenges to doing this in a mixed-use facility has been to quantify the power consumption of each major mechanical or electrical subsystem that is attributable to the IT equipment housed within the ESDC-TB (i.e., NW-ICE). PNNL has tackled this issue for its five chiller plant, and the analysis is presented in the paper. The analysis for all the remaining mechanical and electrical subsystems is now underway and will be presented in future publications.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Accelerating Numerical Calculation on the Cray XMT

The Cray XMT provides hardware support for parallel algorithms that would be communication- or memory-bound on other machines. Unfortunately, even if an algorithm meets these criteria, performance suffers if the algorithm is too numerically intensive. We present a lookup-based approach that achieves a significant performance advantage over explicit calculation. We describe an approach to balancing memory bandwidth against on-chip floating point capabilities, leading to further speedup. Finally, we provide table lookup algorithms for a number of common functions.

97 MATHEMATICS AND COMPUTING↗

Effects of floating-point non-associativity on numerical computations on massively multithreaded systems

Floating-point addition and multiplication are not necessarily associative. When performing those operations over large numbers of operands with different magnitudes, the order in which individual operations are performed can affect the final result. On massively multithreaded systems, when performing parallel reductions, the non-deterministic nature of numerical operation interleaving can lead to non-deterministic numerical results. We have investigated the effect of this problem on the convergence of a conjugate gradient calculation used as part of a power grid analysis application.

97 MATHEMATICS AND COMPUTING↗

Liquid Cooling in Data Centers

Semiconductor manufacturers have aggressively attacked the problem of escalating microprocessor power consumption levels. Today, server manufacturers can purchase microprocessors that currently have power consumption levels capped at 100W maximum. However, total server power levels continue to increase, with the increase in power consumption coming from the supportin chipsets, memory, and other components. In turn, full rack heat loads are very aggressivley climbing as well, and this is making it increasingly difficult and cost-prohibitive for facility owners to cool these high power racks. As a result, facilities owners are turning to alternative, and more energy efficient, cooling solutions that deploy liquids in one form or another. The paper discusses the advent of the adoption of liquid-cooling in high performance computing centers. An overview of the following competing rack-based, liquid-cooling, technologies is provided: in-row, above rack, refrigerated/enclosed rack, rear door heat exchanger, and device-level (i.e., chip-level). Preparation for a liquid-cooled data center, retroft and greenfield (new), is discussed, with a focus on the key issues that are common to all liquid-cooling technologies that depend upon the delivery of water to the rack (or in some deployments, a Coolant Distribution Unit). The paper then discusses, in some detail, the actual implementation and deployment of a liquid device-level cooled (spray cooled) supercomputer at the Pacific Northwest National Laboratory. Initial results from a successful 30 day compliance test show excellent hardware stability, operating system (OS) and software stack stability, application stability and performance, and an availability level that exceeded expectations at 99.94%. The liquid-cooled supercomputer achieved a peak performance of 9.287 TeraFlops, which placed it at number 101 in the June 2007 Top500 fastest supercomputers worldwide. Long-term performance and energy efficiency testing is currently underway, and detailed results will be reported in upcoming publications.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

A parallel WLS state estimator on shared memory computers

This paper describes our experience with developing a parallel weighted- least-square (WLS) state estimation (SE) program for shared-memory parallel computers. Since the key computational kernel of the WLS algorithm based on the Newton-Raphson approach is a solver of sparse linear equations, a significant part of our effort was focused on selecting, implementing and evaluating this algorithm. An optimized shared memory version of the conjugate gradient (CG) algorithm was found to be competitive to state-of-the-art implementation of LU solvers for the SE problem on the SGI Altix, a shared memory architecture. We also ported the full SE algorithm including CG to the Cray MTA-2 shared memory multithreaded architecture and investigated its performance.

97 MATHEMATICS AND COMPUTING↗

PNNL streamlines energy-guzzling computers

In a room the size of a garage, two rows of six-foot-tall racks holding supercomputer hard drives sit back-to-back. Thin tubes and wires snake off the hard drives, slithering into the corners. Stepping between the rows, a rush of heat whips around you -- the air from fans blowing off processing heat. But walk farther in, between the next racks of hard drives, and the temperature drops noticeably. These drives are being cooled by a non-conducting liquid that runs right over the hardworking processors. The liquid carries the heat away in tubes, saving the air a few degrees. This is the Energy Smart Data Center at Pacific Northwest National Laboratory. The bigger, faster, and meatier supercomputers get, the more energy they consume. PNNL's Andres Marquez has developed this test bed to learn how to train the behemoths in energy efficiency. The work will help supercomputers perform better as well. Processors have to keep cool or suffer from "thermal throttling," says Marquez. "That's the performance threshold where the computer is too hot to run well. That threshold is an industry secret." The center at EMSL, DOE's national scientific user facility at PNNL, harbors several ways of experimenting with energy usage. For example, the room's air conditioning is isolated from the rest of EMSL -- pipes running beneath the floor carry temperature-controlled water through heat exchangers to cooling towers outside. "We can test whether it's more energy efficient to cool directly on the processing chips or out in the water tower," says Marquez. The hard drives feed energy and temperature data to a network server running specially designed software that controls and monitors the data center. To test the center’s limits, the team runs the processors flat out – not only on carefully controlled test programs in the Energy Smart computers, but also on real world software from other EMSL research, such as regional weather forecasting models. Marquez's group is also developing "power aware computing", where the computer programs themselves perform calculations more energy efficiently. Maybe once computers get smart about energy, they'll have tips for their users.

data center↗