Date:\u00a0 April 25 – 26, 2019<\/p>\n
Place: Department of Computer Science<\/a>,\u00a0U<\/span>iT The Arctic University of Norway<\/span><\/a>, Tromso, Norway<\/p>\n Room:\u00a0Technology building, room\u00a01.023<\/a> (for keynote 1) and\u00a0room 3.028<\/a>.<\/p>\n Large-scale analytics based on big data is crucial for scientific discovery and societal digitalization but poses great challenges on sustainable data processing. In data-intensive science, projected growth rate (CAGR) for data acquisition is 72%, two times CAGR for processor capacity (36%) and more than three times CAGR for memory capacity (20%). In societal services such as web search, a single<\/em> web search touches more than fifty separate internal services and thousands of machines. Google data centers consume almost 260 MW, about a quarter of the output of a nuclear power plant, enough to power 200 000 homes. Therefore, data analytics systems that will be sustainable over the data deluge, must be resource-efficient.<\/p>\n Efficient computing technologies that have been evolved over decades in high performance computing (HPC) are the key enablers to address the big data processing challenges. While conventional HPC systems focus on simulating physical systems, new computing systems for analyzing a large amount of data have recently emerged. Today scientific discovery typically involves both HPC and big data analytics (BDA), demanding the integration of HPC and BDA. The integration will not only address the current big data challenges but also enable new scientific discovery. To achieve the integration of HPC and BDA, it is crucial to unify their currently divergent computing methodologies.<\/p>\n The workshop will discuss UPC\/UPC++, HPC technologies, data-intensive scientific applications and the possibilities of leveraging UPC\/UPC++ and\u00a0 HPC technologies for data-intensive applications. Invited speakers from Lawrence Berkeley National Laboratory – USA, Simula Laboratory – Norway and UiT The Arctic University of Norway will be giving keynote presentations on the subject.<\/p>\n 09:00 – 10:00 Keynote 1: The UPC++ Library for Exascale and Data Intensive Computing\u00a0<\/strong>(slides<\/a>)<\/p>\n Speaker: Prof.\u00a0Scott B. Baden, Group Lead, Computer Languages and System Software Group, Computational Research Division, Lawrence Berkeley National Laboratory<\/p>\n Place:\u00a0\u00a0Room 1.023, Teknologibygget<\/a><\/strong><\/p>\n Abstract: UPC++ is a C++ library that supports high-performance computation via an asynchronous communication framework. UPC++ runs under the PGAS execution model and supports applications where communication is irregular, fine-grained, or both.<\/p>\n UPC++’s combination of low-overhead, one-sided communication, remote procedure call, and aggressive asynchrony is the key to delivering high performance with irregular applications and enabling improved programmer productivity. UPC++ futures enable the programmer to capture data readiness state. They may be composed and synchronized in bulk, expressing dependence-driven execution of aggregated operations without the need for explicit busy waiting. UPC++ has recently added RMA support for GPU memory, so that data may be moved transparently between memories with different optimal access methods. The feature, called memory kinds, is the basis for extending UPC++ to data intensive applications.<\/p>\n The UPC++ programmer can expect communication to run at close to hardware speeds, thanks to support from the GASNet-EX library, which and leverages low-overhead communication and access to special hardware support, e.g. RDMA.<\/p>\n I will present examples that demonstrate the elegance of UPC++’s abstractions together with performance results obtained on scalable systems.<\/p>\n Bio: Dr. Baden received his M.S and Ph.D. in Computer Science from UC Berkeley in 1982 and 1987. He is also Adjunct Professor in the Department of Computer Science and Engineering at UCSD, where he was a member of the regular faculty for 27 years. His research interests are in high performance and scientific computation: domain-specific translation, abstraction mechanisms, run times, and irregular problems.<\/p>\n <\/p>\n 10:15 – 11:00\u00a0<\/strong>Keynote 2:\u00a0Compute- and Data-Intensive Aspects of Computational Nanoscopy <\/strong>(slides<\/a>)<\/p>\n Speaker: Prof.\u00a0\u00a0Krishna Agarwal, ERC grantee, UiT The Arctic University of Norway<\/p>\n Abstract:\u00a0Nanoscopy has provided a window to the world of sub-cellular biology, supporting 20-100 nm computational resolution instead of microscopy\u2019s 250 nm optical resolution. At the same time, microscope and nanoscopy have scaled to provide large field-of-view imaging, 3D imaging, and 4D imaging. Although the scale of observation or dimensionality of imaging may increase linearly, the data and compute requirements increase exponentially. Even when the field of microscopy and nanoscopy is on the verge of creating large scale impact on the society, the main bottleneck now is data and compute bottleneck. In this talk, I discuss the urgency of this situation and how data and compute scientists can participate in resolving the bottleneck and contribute to the significant impact of this field.<\/p>\n <\/p>\n 11:15 – 12:00\u00a0Unified Frameworks for\u00a0Data\u2010Intensive and Compute\u2010Intensive Applications <\/strong>(slides<\/a>)<\/p>\n Speaker: Dr. Amin Khan,\u00a0UiT The Arctic University of Norway<\/p>\n Abstract: Building data-intensive and compute-intensive applications requires frameworks and systems that can achieve high performance dealing efficiently with the unparalleled rise in the scale of data and computation with the arrival of exascale computing. At the same time, these frameworks needs to be expressive and intuitively programmable, in order to limit the effort and costs involved in the application development and maintenance. In this talk, we explore the recent developments and future trends in the state-of-the-art to build big data processing frameworks that strive to achieve programmability and productivity of big data platforms like Hadoop and Spark, while at the same time keeping pace with the performance of HPC systems.<\/p>\n <\/p>\n 12:30 – 14:00 Lunch. <\/strong>Place:Pharmacy canteen<\/p>\n <\/p>\n 14:15 – 15:00<\/strong>\u00a0Keynote 3: PGAS for graph analytics: can one sided communications break the scalability barrier? <\/strong>(slides<\/a>)<\/p>\n Speaker: Dr.\u00a0Johannes Langguth, Research scientist, Simula Laboratory, Norway<\/p>\n Abstract:\u00a0As the world is becoming increasingly interconnected and systems increasingly complex. Therefore, technologies that can analyze connected systems and their dynamic characteristics become indispensable. Consequently, the last decade has seen increasing interest in graph analytics, which allows obtaining insights from such connected data. Parallel graph analytics can reveal the workings of intricate systems and networks at massive scales, which are found in diverse areas such as social networks, economic transactions, and protein interactions. While sequential graph algorithms have been studied for decades, the recent availability of massive datasets has given rise to the need for parallel graph processing, which poses unique challenges.<\/p>\n Benchmarks such as the Graph 500 have shown that graph processing performance is largely unrelated to traditional measurements of performance such as FLOPS or memory bandwidth. Instead, algorithmic communication aggregation and network latencies play a crucial role here.<\/p>\n In this talk we introduce the area of parallel graph analytics with a special focus on news dissemination, along with the technical challenges it presents and discuss how PGAS systems with support for one-sided messaging, such as UPC++, can help in overcoming these challenges.<\/p>\n <\/p>\n 15:15 – 16:00\u00a0Efficient Communication for Data-Intensive Applications <\/strong>(slides<\/a>)<\/p>\n Speaker: Dr. Nga Dinh,\u00a0UiT The Arctic University of Norway<\/p>\n <\/p>\n 16:15 – 17:00\u00a0Leveraging High-Performance Computing (HPC) for Big Data Analytics (BDA) <\/strong>(slides<\/a>)<\/p>\n Speaker: MSc. Ngoc Phuong Chau, UiT The Arctic University of Norway<\/p>\n <\/p>\n 17:00 – 18<\/strong>:00 Discussion<\/strong><\/p>\n 19:00 – 21:30 Dinner \/ Social event<\/strong><\/p>\n <\/p>\n 09:15 – 10:00 Keynote 4:\u00a0Heterogeneous Computing for Cardiac Electrophysiology <\/span><\/strong>(slides<\/a>)<\/span><\/p>\n Speaker: Prof. Xing Cai, Chief Research Scientist, Simula Laboratory, Norway<\/p>\n Abstract:\u00a0Electrical activities inside the heart are immensely important for the functioning of this vital organ. In the pursuit of a scientific understanding of the processes and mechanisms in electro-physiology, computer\u00a0simulations have become an established paradigm of research. Both the complex mathematical models and the extreme\u00a0physiological details require huge-scale simulations, which nowadays see an increasing use of heterogeneous computing. That is, the computational power is delivered by more than one\u00a0processor type. We will discuss some of the resulting challenges in programming and performance\u00a0optimization. Successful applications from the domain of cardiac\u00a0electro-physiology will be used to demonstrate\u00a0the usefulness of\u00a0heterogeneous computing. We will also take a peek into the future of heterogeneous computing through eX3: the brand-new national\u00a0infrastructure for experimental exploration of exascale computing.<\/p>\n <\/p>\n 10:15 – 11:00\u00a0Keynote 5:\u00a0 Scalable Machine Learning:\u00a0Practical situational awareness for autonomous maritime navigation – Are we there? <\/strong>(slides<\/a>)<\/p>\n Speaker: Prof.\u00a0 Dilip Prasad, UiT The Arctic University of Norway<\/p>\n Abstract: Autonomous maritime navigation requires autonomous situational awareness. For this, intelligible knowledge has to emerge from multiple and diverse sensors. Even a subset of maritime sensors, such as electro-optical cameras, can generate data and information-heavy video streams. The machine learning approaches for deriving information from data are very challenging for maritime scenario. This is because the number and variety of objects entering the navigation scene can vary dramatically and there may be no landmarks to utilize in maritime scenes. The learning has to be done incrementally, should be scalable for varying number of objects, and adaptable to their varying motion patterns. Do the current algorithms cater to these needs? Are the current algorithms adaptable to include more number of variety of asynchronous sensors?<\/p>\n <\/p>\n 11:15 – 12:00\u00a0Energy-efficient in-situ analytics <\/strong>(slides<\/a>)<\/p>\n Speaker: MSc. Cheng-Hsiang Chiu, UiT The Arctic University of Norway<\/p>\n <\/p>\n 12:30 – 17:30 Excursion with lunch<\/strong><\/p>\n 17:30 – 18:00 Closing<\/strong><\/p>\n <\/p>\n
\n
\nAgenda (tentative)<\/h1>\n
Thursday, Apr. 25, 2019<\/h2>\n
<\/h2>\n
Friday, Apr. 26, 2019<\/h2>\n