Search NASA⌕ Search

SEARCH · Search NASA

Results for “Offloading”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 397 records · Page 22

Overlapping Computation and Communication: Barrier Algorithms and ConnectX-2 CORE-Direct Capabilities

This paper explores the computation and communication overlap capabilities enabled by the new CORE-Direct hardware capabilities introduced in the InfiniBand (IB) Host Channel Adapter (HCA) ConnectX-2. These capabilities enable the progression and completion of data-dependent communications sequences to progress and complete at the network level without any Central Processing Unit (CPU) involvement. We use the latency dominated nonblocking barrier algorithm in this study, and find that at 64 process count, a contiguous time slot of about 80 percent of the nonblocking barrier time is available for computation. This time slot increases as the number of processes participating increases. In contrast, CPU based implementations provide a time slot of up to 30 percent of the nonblocking barrier time. This bodes well for the scalability of simulations employing offloaded collective operations These capabilities can be used to reduce the effects of system noise, and when using nonblocking collective operations have the potential to hide the effects of application load imbalance.

Graham, Richard L↗

ConnectX-2 CORE-Direct Enabled Asynchronous Broadcast Collective Communications

This paper describes the design and implementation of InfiniBand (IB) CORE-Direct based blocking and nonblocking broadcast operations within the Cheetah collective operation framework. It describes a novel approach that fully ofFLoads collective operations and employs only user-supplied buffers. For a 64 rank communicator, the latency of CORE-Direct based hierarchical algorithm is better than production-grade Message Passing Interface (MPI) implementations, 150% better than the default Open MPI algorithm and 115% better than the shared memory optimized MVAPICH implementation for a one kilobyte (KB) message, and for eight mega-bytes (MB) it is 48% and 64% better, respectively. Flat-topology broadcast achieves 99.9% overlap in a polling based communication-computation test, and 95.1% overlap for a wait based test, compared with 92.4% and 17.0%, respectively, for a similar Central Processing Unit (CPU) based implementation.

99 GENERAL AND MISCELLANEOUS↗

CATCH: A Cloud-based Adaptive Data Transfer Service for HPC

Modern High Performance Computing (HPC) applications process very large amounts of data. A critical research challenge lies in transporting input data to the HPC center from a number of distributed sources, e.g., scientific experiments and web repositories, etc., and offloading the result data to geographically distributed, intermittently available end-users, often over under-provisioned connections. Such end-user data services are typically performed using point-to-point transfers that are designed for well-endowed sites and are unable to reconcile the center's resource usage and users delivery deadlines, unable to adapt to changing dynamics in the end-to-end data path and are not fault-tolerant. To overcome these inefficiencies, decentralized HPC data services are emerging as viable alternatives. In this paper, we develop and enhance such distributed data services by designing CATCH, a Cloud-based Adaptive data Transfer serviCe for HPC. CATCH leverages a bevy of cloud storage resources to orchestrate a decentralized data transport with fail-over capabilities. Our results demonstrate that CATCH is a feasible approach, and can help improve the data transfer times at the HPC center by as much as 81.1% for typical HPC workloads.

99 GENERAL AND MISCELLANEOUS↗

A highly reliable RAID system based on GPUs.

While RAID is the prevailing method of creating reliable secondary storage infrastructure, many users desire more flexibility than offered by current implementations. To attain needed performance, customers have often sought after hardware-based RAID solutions. This talk describes a RAID system that offloads erasure correction coding calculations to GPUs, allowing increased reliability by supporting new RAID levels while maintaining high performance.

99 GENERAL AND MISCELLANEOUS↗

Scalable Memory Registration for High-Performance Networks Using Helper Threads

Remote DMA (RDMA) enables high performance networks to reduce data copying between an application and the operating system (OS). However RDMA operations in some high performance networks require communication memory explicitly registered with the network adapter and pinned by the OS. Memory registration and pinning limits the flexibility of the memory system and reduces the amount of memory that user processes can allocate. These issues become more significant on multicore platforms, since registered memory demand grows linearly with the number of processor cores. In this paper we propose a new memory registration/deregistration strategy to reduce registered memory on multicore architectures for HPC applications. We hide the cost of dynamic memory management by offloading all dynamic memory registration and deregistration requests to a dedicated memory management helper thread. We investigate design policies and performance implications of the helper thread approach. We evaluate our framework with the NAS parallel benchmarks, for which our registration scheme significantly reduces the registered memory (23.62% on average and up to 49.39%) and avoids memory registration/deregistration costs for reused communication memory. We show that our system enables the execution of problem sizes that could not complete under existing memory registration strategies.

99 GENERAL AND MISCELLANEOUS↗

Distributed Storage Systems for Data Intensive Computing

In this chapter, the authors present an overview of the utility of distributed storage systems in supporting modern applications that are increasingly becoming data intensive. Their coverage of distributed storage systems is based on the requirements imposed by data intensive computing and not a mere summary of storage systems. To this end, they delve into several aspects of supporting data-intensive analysis, such as data staging, offloading, checkpointing, and end-user access to terabytes of data, and illustrate the use of novel techniques and methodologies for realizing distributed storage systems therein. The data deluge from scientific experiments, observations, and simulations is affecting all of the aforementioned day-to-day operations in data-intensive computing. Modern distributed storage systems employ techniques that can help improve application performance, alleviate I/O bandwidth bottleneck, mask failures, and improve data availability. They present key guiding principles involved in the construction of such storage systems, associated tradeoffs, design, and architecture, all with an eye toward addressing challenges of data-intensive scientific applications. They highlight the concepts involved using several case studies of state-of-the-art storage systems that are currently available in the data-intensive computing landscape.

99 GENERAL AND MISCELLANEOUS↗

Active Flash: Performance-Energy Tradeoffs for Out-of-Core Processing on Non-Volatile Memory Devices

In this abstract, we study the performance and energy tradeoffs involved in migrating data analysis into the flash device, a process we refer to as Active Flash. The Active Flash paradigm is similar to 'active disks', which has received considerable attention. Active Flash allows us to move processing closer to data, thereby minimizing data movement costs and reducing power consumption. It enables true out-of-core computation. The conventional definition of out-of-core solvers refers to an approach to process data that is too large to fit in the main memory and, consequently, requires access to disk. However, in Active Flash, processing outside the host CPU literally frees the core and achieves real 'out-of-core' analysis. Moving analysis to data has long been desirable, not just at this level, but at all levels of the system hierarchy. However, this requires a detailed study on the tradeoffs involved in achieving analysis turnaround under an acceptable energy envelope. To this end, we first need to evaluate if there is enough computing power on the flash device to warrant such an exploration. Flash processors require decent computing power to run the internal logic pertaining to the Flash Translation Layer (FTL), which is responsible for operations such as address translation, garbage collection (GC) and wear-leveling. Modern SSDs are composed of multiple packages and several flash chips within a package. The packages are connected using multiple I/O channels to offer high I/O bandwidth. SSD computing power is also expected to be high enough to exploit such inherent internal parallelism within the drive to increase the bandwidth and to handle fast I/O requests. More recently, SSD devices are being equipped with powerful processing units and are even embedded with multicore CPUs (e.g. ARM Cortex-A9 embedded processor is advertised to reach 2GHz frequency and deliver 5000 DMIPS; OCZ RevoDrive X2 SSD has 4 SandForce controllers, each with 780MHz max frequency Tensilica core). Efforts that take advantage of the available computing cycles on the processors on SSDs to run auxiliary tasks other than actual I/O requests are beginning to emerge. Kim et al. investigate database scan operations in the context of processing on the SSDs, and propose dedicated hardware logic to speed up scans. Also, cluster architectures have been explored, which consist of low-power embedded CPUs coupled with small local flash to achieve fast, parallel access to data. Processor utilization on SSD is highly dependent on workloads and, therefore, they can be idle during periods with no I/O accesses. We propose to use the available processing capability on the SSD to run tasks that can be offloaded from the host. This paper makes the following contributions: (1) We have investigated Active Flash and its potential to optimize the total energy cost, including power consumption on the host and the flash device; (2) We have developed analytical models to analyze the performance-energy tradeoffs for Active Flash, by treating the SSD as a blackbox, this is particularly valuable due to the proprietary nature of the SSD internal hardware; and (3) We have enhanced a well-known SSD simulator (from MSR) to implement 'on-the-fly' data compression using Active Flash. Our results provide a window into striking a balance between energy consumption and application performance.

42 ENGINEERING↗

Active Flash: Out-of-core Data Analytics on Flash Storage

Next generation science will increasingly come to rely on the ability to perform efficient, on-the-fly analytics of data generated by high-performance computing (HPC) simulations, modeling complex physical phenomena. Scientific computing workflows are stymied by the traditional chaining of simulation and data analysis, creating multiple rounds of redundant reads and writes to the storage system, which grows in cost with the ever-increasing gap between compute and storage speeds in HPC clusters. Recent HPC acquisitions have introduced compute node-local flash storage as a means to alleviate this I/O bottleneck. We propose a novel approach, Active Flash, to expedite data analysis pipelines by migrating to the location of the data, the flash device itself. We argue that Active Flash has the potential to enable true out-of-core data analytics by freeing up both the compute core and the associated main memory. By performing analysis locally, dependence on limited bandwidth to a central storage system is reduced, while allowing this analysis to proceed in parallel with the main application. In addition, offloading work from the host to the more power-efficient controller reduces peak system power usage, which is already in the megawatt range and poses a major barrier to HPC system scalability. We propose an architecture for Active Flash, explore energy and performance trade-offs in moving computation from host to storage, demonstrate the ability of appropriate embedded controllers to perform data analysis and reduction tasks at speeds sufficient for this application, and present a simulation study of Active Flash scheduling policies. These results show the viability of the Active Flash model, and its capability to potentially have a transformative impact on scientific data analysis.

99 GENERAL AND MISCELLANEOUS↗

A report documenting the completion of the Los Alamos National Laboratory portion of the ASC level II milestone ""Visualization on the supercomputing platform

This report provides documentation for the completion of the Los Alamos portion of the ASC Level II 'Visualization on the Supercomputing Platform' milestone. This ASC Level II milestone is a joint milestone between Sandia National Laboratory and Los Alamos National Laboratory. The milestone text is shown in Figure 1 with the Los Alamos portions highlighted in boldfaced text. Visualization and analysis of petascale data is limited by several factors which must be addressed as ACES delivers the Cielo platform. Two primary difficulties are: (1) Performance of interactive rendering, which is the most computationally intensive portion of the visualization process. For terascale platforms, commodity clusters with graphics processors (GPUs) have been used for interactive rendering. For petascale platforms, visualization and rendering may be able to run efficiently on the supercomputer platform itself. (2) I/O bandwidth, which limits how much information can be written to disk. If we simply analyze the sparse information that is saved to disk we miss the opportunity to analyze the rich information produced every timestep by the simulation. For the first issue, we are pursuing in-situ analysis, in which simulations are coupled directly with analysis libraries at runtime. This milestone will evaluate the visualization and rendering performance of current and next generation supercomputers in contrast to GPU-based visualization clusters, and evaluate the perfromance of common analysis libraries coupled with the simulation that analyze and write data to disk during a running simulation. This milestone will explore, evaluate and advance the maturity level of these technologies and their applicability to problems of interest to the ASC program. In conclusion, we improved CPU-based rendering performance by a a factor of 2-10 times on our tests. In addition, we evaluated CPU and CPU-based rendering performance. We encourage production visualization experts to consider using CPU-based rendering solutions when it is appropriate. For example, on remote supercomputers CPU-based rendering can offer a means of viewing data without having to offload the data or geometry onto a CPU-based visualization system. In terms of comparative performance of the CPU and CPU we believe that further optimizations of the performance of both CPU or CPU-based rendering are possible. The simulation community is currently confronting this reality as they work to port their simulations to different hardware architectures. What is interesting about CPU rendering of massive datasets is that for part two decades CPU performance has significantly outperformed CPU-based systems. Based on our advancements, evaluations and explorations we believe that CPU-based rendering has returned as one viable option for the visualization of massive datasets.

97 MATHEMATICS AND COMPUTING↗

Assessing the Performance and Scalability of a Novel Multilevel K-Nomial Allgather on CORE-Direct Systems

In this paper, we propose a novel allgather algorithm, Reindexed Recursive K-ing (RRK), which leverages flexibility in the algorithm's tree topology and ability to make asynchronous progress coupled with Core-Direct communication offload capability to optimize the MPI-Allgather for Core-Direct enabled systems. In particular, the RRK introduces a reindexing scheme which ensures contiguous data transfers while adding only a single additional send and receive operation for any radix, k, or communicator size, N. This allows us to improve algorithm scalability by avoiding the use of a scatter/gather elements (SGE) list on InfiniBand networks. The implementations of the RRK algorithm and its evaluation shows that it performs and scales well on Core-Direct systems for a wide range of message sizes and various communicator configurations.

99 GENERAL AND MISCELLANEOUS↗

A scalable messaging system for accelerating discovery from large scale scientific simulations

Emerging scientific and engineering simulations running at scale on leadership-class High End Computing (HEC) environments are producing large volumes of data, which has to be transported and analyzed before any insights can result from these simulations. The complexity and cost (in terms of time and energy) associated with managing and analyzing this data have become significant challenges, and are limiting the impact of these simulations. Recently, data-staging approaches along with in-situ and in-transit analytics have been proposed to address these challenges by offloading I/O and/or moving data processing closer to the data. However, scientists continue to be overwhelmed by the large data volumes and data rates. In this paper we address this latter challenge. Specifically, we propose a highly scalable and low-overhead associative messaging framework that runs on the data staging resources within the HEC platform, and builds on the staging-based online in-situ/in- transit analytics to provide publish/subscribe/notification-type messaging patterns to the scientist. Rather than having to ingest and inspect the data volumes, this messaging system allows scientists to (1) dynamically subscribe to data events of interest, e.g., simple data values or a complex function or simple reduction (max()/min()/avg()) of the data values in a certain region of the application domain is greater/less than a threshold value, or certain spatial/temporal data features or data patterns are detected; (2) define customized in-situ/in-transit actions that are triggered based on the events, such as data visualization or transformation; and (3) get notified when these events occur. The key contribution of this paper is a design and implementation that can support such a messaging abstraction at scale on high- end computing (HEC) systems with minimal overheads. We have implemented and deployed the messaging system on the Jaguar Cray XK6 machines at Oak Ridge National Laboratory and the Lonestar system at the Texas Advanced Computing Center (TACC), and we present the experimental performance evaluation using these HEC platforms in the paper.

Jin, Tong↗

Systems and methods for managing shared-path instrumentation and irradiation targets in a nuclear reactor

Systems and methods permit discriminate access to nuclear reactors. Systems provide penetration pathways to irradiation target loading and offloading systems, instrumentation systems, and other external systems at desired times, while limiting such access during undesired times. Systems use selection mechanisms that can be strategically positioned for space sharing to connect only desired systems to a reactor. Selection mechanisms include distinct paths, forks, diverters, turntables, and other types of selectors. Management methods with such systems permits use of the nuclear reactor and penetration pathways between different systems and functions, simultaneously and at only distinct desired times. Existing TIP drives and other known instrumentation and plant systems are useable with access management systems and methods, which can be used in any nuclear plant with access restrictions.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Quantum Accelerators for High-performance Computing Systems

We define some of the programming and system-level challenges facing the application of quantum processing to high-performance computing. Alongside barriers to physical integration, prominent differences in the execution of quantum and conventional programs challenges the intersection of these computational models. Following a brief overview of the state of the art, we discuss recent advances in programming and execution models for hybrid quantum-classical computing. We discuss a novel quantum-accelerator framework that uses specialized kernels to offload select workloads while integrating with existing computing infrastructure. We elaborate on the role of the host operating system to manage these unique accelerator resources, the prospects for deploying quantum modules, and the requirements placed on the language hierarchy connecting these different system components. We draw on recent advances in the modeling and simulation of quantum computing systems with the development of architectures for hybrid high-performance computing systems and the realization of software stacks for controlling quantum devices. Finally, we present simulation results that describe the expected system-level behavior of high-performance computing systems composed from compute nodes with quantum processing units. We describe performance for these hybrid systems in terms of time-to-solution, accuracy, and energy consumption, and we use simple application examples to estimate the performance advantage of quantum acceleration.

Humble, Travis↗

STAR'S Approach to Highly Efficient End-To-End Grid Production

STAR's RHIC computing facility provides over 15K dedicated slots for data reconstruction. This number of slots is not always sufficient to satisfy an ambitious and data-challenging physics program, and harvesting resources from outside facilities is paramount to scientific success. However, constraints of remote sites (e.g. CPU time limits) do not always provide the flexibility of a dedicated farm. STAR has a breadth of smaller datasets (both in runtime and size) that can be easily offloaded to remote facilities with many such limits. Scavenged resources can be run with efficiency comparable to that of the local production and contributes additional computing time to an experiment that runs every year and therefore needs fast turnaround. We will discuss STAR's software stack of our grid production framework including features dealing with multi-site submission, automated re-submission, job tracking, as well as new challenges and possible improvements.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Packet capture engine for commodity network interface cards in high-speed networks

A method, systems for a packet capture engine for commodity network interface cards (NICs) in high-speed networks that provides lossless zero-copy packet capture and delivery services by exploiting multi-queue NICs and multicore architectures. The methods, systems include a ring-buffer-pool mechanism and a buddy-group based offloading mechanism.

97 MATHEMATICS AND COMPUTING↗

The Castro AMR Simulation Code: Current and Future Developments

We describe recent developments to the Castro astrophysics simulation code, focusing on new features that enable our simulations of X-ray bursts. Two highlights of Castro's ongoing development are the new integration technique to couple hydrodynamics and reactions to high order and GPU offloading. We discuss how these features will help offset some of the computational expense in X-ray burst models.

79 ASTRONOMY AND ASTROPHYSICS↗

CCAMP: An Integrated Translation and Optimization Framework for OpenACC and OpenMP

Heterogeneous computing and exploration into specialized accelerators are inevitable in current and future supercomputers. Although this diversity of devices is promising for performance, the array of architectures presents programming challenges. High-level programming strategies have emerged to face these challenges, such as the OpenMP offloading model and OpenACC. The varying levels of support for these standards, however, within vendor-specific and open-source tools, as well as the lack of performance portability across devices, have prevented the standards from achieving their goals. To address these shortcomings, we present CCAMP, an OpenMP and OpenACC interoperable framework. CCAMP provides two primary facilities: language translation between the two standards and device-specific directive optimization within each standard. We show that by using the CCAMP framework, programmers can easily transplant non-portable code into new ecosystems for new architectures. Additionally, by using CCAMP device-specific directive optimizations, users can achieve optimized performance across architectures using a single source code.

Lambert, Jacob↗

GPU Usage in ATLAS Reconstruction and Analysis

With Graphical Processing Units (GPUs) and other kinds of accelerators becoming ever more accessible, High Performance Computing Centres all around the world using them ever more, ATLAS has to find the best way of making use of such accelerators in much of its computing. Tests with GPUs - mainly with CUDA - have been performed in the past in the experiment. At that time the conclusion was that it was not advantageous for the ATLAS offline and trigger software to invest time and money into GPUs. However as the usage of accelerators has become cheaper and simpler in recent years, their re-evaluation in ATLAS's offline software is warranted. We show new results of using GPU accelerated calculations in ATLAS's offline software environment using the ATLAS offline/analysis (xAOD) Event Data Model. We compare the performance and flexibility of a couple of the available GPU programming methods, and show how different memory management setups affect our ability to offload different types of calculations to a GPU efficiently.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗