MPI tools and performance studies---MPI performance analysis tools on Blue Gene/L

Chung, I-Hsin; Walkup, R. E.; Wen, Hui-Fang; Yu, Hao

doi:10.1145/1188455.1188583

Cited by 24 publications

(28 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With the growing popularity and complexity of parallel programming models, visualizations of communication behavior has been included into several performance tools [9]. One of the most frequently used visualizations is that of traces of MPI messages sent.…”

Section: Communication Analysis To Improve Performancementioning

confidence: 99%

Visualizing Network Traffic to Understand the Performance of Massively Parallel Simulations

Landge

Levine

Bhatelé

et al. 2012

IEEE Trans. Visual. Comput. Graphics

View full text Add to dashboard Cite

Fig. 1. Network traffic resulting from two different runs of the parallel simulation pF3D. This simulation models laser plasma interaction inside of a hohlraum chamber by decomposing the domain into a set of blocks (left). Depending on how data blocks are mapped to processor cores (middle), different communication patterns occur. When staggering data placement (bottom right) we observe significantly more balanced communication compared to a default mapping similar to how the domain is decomposed (top right).Abstract-The performance of massively parallel applications is often heavily impacted by the cost of communication among compute nodes. However, determining how to best use the network is a formidable task, made challenging by the ever increasing size and complexity of modern supercomputers. This paper applies visualization techniques to aid parallel application developers in understanding the network activity by enabling a detailed exploration of the flow of packets through the hardware interconnect. In order to visualize this large and complex data, we employ two linked views of the hardware network. The first is a 2D view, that represents the network structure as one of several simplified planar projections. This view is designed to allow a user to easily identify trends and patterns in the network traffic. The second is a 3D view that augments the 2D view by preserving the physical network topology and providing a context that is familiar to the application developers. Using the massively parallel multi-physics code pF3D as a case study, we demonstrate that our tool provides valuable insight that we use to explain and optimize pF3D's performance on an IBM Blue Gene/P system.

show abstract

Section: Communication Analysis To Improve Performancementioning

confidence: 99%

Visualizing Network Traffic to Understand the Performance of Massively Parallel Simulations

Landge

Levine

Bhatelé

et al. 2012

IEEE Trans. Visual. Comput. Graphics

View full text Add to dashboard Cite

show abstract

“…The techniques vary with the different network topologies. Specifically, mapping optimizations for Blue Gene torus networks [12], [13] take the application communication logs as an input and generate mapfiles for future application runs to optimize the hop-byte metric. Bhatele et al [12] also show benefits in a single domain WRF run with their techniques.…”

Section: Related Workmentioning

confidence: 99%

“…For example, in WRF, each integration time-step involves 144 message exchanges with the four neighbouring processes [3]. IBM's HPCT [21] profiling tools show that about 40% of the total execution time in WRF is spent in communication.…”

Section: Mappingmentioning

confidence: 99%

A divide and conquer strategy for scaling weather simulations with multiple regions of interest

Malakar¹,

George²,

Kumar³

et al. 2012

2012 International Conference for High Performance Computing, Networking, Storage and Analysis

View full text Add to dashboard Cite

Abstract-Accurate and timely prediction of weather phenomena, such as hurricanes and flash floods, require highfidelity compute intensive simulations of multiple finer regions of interest within a coarse simulation domain. Current weather applications execute these nested simulations sequentially using all the available processors, which is sub-optimal due to their sublinear scalability. In this work, we present a strategy for parallel execution of multiple nested domain simulations based on partitioning the 2-D processor grid into disjoint rectangular regions associated with each domain. We propose a novel combination of performance prediction, processor allocation methods and topology-aware mapping of the regions on torus interconnects. Experiments on IBM Blue Gene systems using WRF show that the proposed strategies result in performance improvement of up to 33% with topology-oblivious mapping and up to additional 7% with topology-aware mapping over the default sequential strategy.

show abstract

“…Chung et al [20] evaluated several of the aforementioned MPI performance analysis tools on BlueGene/L and identified the quantity of data collected at large scales as a potential bottleneck. Schulz et al [21] described an implementation of the Dynamic Probe Class Library [22] targeting BlueGene/L.…”

Section: Related Workmentioning

confidence: 99%

Lessons learned at 208K: Towards debugging millions of cores

L¹,

Ahn²,

Arnold³

et al. 2008

2008 SC - International Conference for High Performance Computing, Networking, Storage and Analysis

View full text Add to dashboard Cite

Abstract-Petascale systems will present several new challenges to performance and correctness tools. Such machines may contain millions of cores, requiring that tools use scalable data structures and analysis algorithms to collect and to process application data. In addition, at such scales, each tool itself will become a large parallel application -already, debugging the full BlueGene/L (BG/L) installation at the Lawrence Livermore National Laboratory requires employing 1664 tool daemons. To reach such sizes and beyond, tools must use a scalable communication infrastructure and manage their own tool processes efficiently. Some system resources, such as the file system, may also become tool bottlenecks.In this paper, we present challenges to petascale tool development, using the Stack Trace Analysis Tool (STAT) as a case study. STAT is a lightweight tool that gathers and merges stack traces from a parallel application to identify process equivalence classes. We use results gathered at thousands of tasks on an Infiniband cluster and results up to 208K processes on BG/L to identify current scalability issues as well as challenges that will be faced at the petascale. We then present implemented solutions to these challenges and show the resulting performance improvements. We also discuss future plans to meet the debugging demands of petascale machines.

show abstract

MPI tools and performance studies---MPI performance analysis tools on Blue Gene/L

Cited by 24 publications

References 42 publications

Visualizing Network Traffic to Understand the Performance of Massively Parallel Simulations

Visualizing Network Traffic to Understand the Performance of Massively Parallel Simulations

A divide and conquer strategy for scaling weather simulations with multiple regions of interest

Lessons learned at 208K: Towards debugging millions of cores

Contact Info

Product

Resources

About