HPC Wire

Subscribe to HPC Wire feed
Since 1987 - Covering the Fastest Computers in the World and the People Who Run Them
Updated: 6 hours 5 min ago

U of Illinois, NCSA Launch First US Nanomanufacturing Node

Thu, 09/14/2017 - 17:49

The University of Illinois at Urbana-Champaign together with the National Center for Supercomputing Applications (NCSA) have launched the United States’s first computational node aimed at the development of nanomanufacturing simulation tools. NanoMFG Node, as the five-year project is called, began on September 1, 2017, funded by a $4 million grant from the National Science Foundation (NSF).

As described in the NSF award documents:

The mission of the nanoMFG Node is to be the engine for design, simulation, planning, and optimization of highly relevant nano-manufacturing growth and patterning processes. To help achieve this mission, computational tools for nanomanufacturing aimed at multiscale theory, modeling, and simulation (TM&S) will be developed and broadly disseminated. The intellectual merit of this 5-year activity is the development of nanomanufacturing simulation tools, validated by experimental data, and integrated with data-driven uncertainty quantification.

The framework for the effort is based on a layered computational tool infrastructure comprising the creation of the following: (1) nanoscale transport phenomena models, (2) process models, (3) uncertainty quantification framework, (4) application and empirical validation of process models, (5) tools for multiscale transport phenomena, and (6) tools for nanoscale self-assembly.

MechSE professors Placid Ferreira, Kimani Toussaint, Narayana Aluru, and Elif Ertekin.

The tools developed by the node will be validated by experimental data and made available on the nanoHUB, the cyber platform for the Network for Computational Nanotechnology (NCN). A successful endeavor will produce nanomanufacturing simulation tools that improve efficiencies and strengthen the economic viability of nanomanufacturing.

The nanoMFG Node team includes faculty members from the Department of Mechanical Science and Engineering (MechSE) at Illinois — Kimani Toussaint (PI and Director), NCSA’s Narayana Aluru (Co-PI), Elif Ertekin (Co-PI), and Placid Ferreira (Co-PI) — and Hayden Taylor (Co-PI) from UC, Berkeley.

“To make nanomanufacturing economically viable, we envision end-users getting onto the nanoHUB cyber platform and simulating every stage in the manufacturing of a nano-enabled product,” Toussaint said in an official statement. “These simulation tools could save significant time and money while providing valuable insight on how to refine critical process steps in nanomanufacturing.”

The researchers will collaborate with many of the facilities and centers at Illinois, including the Materials Research Laboratory and the Micro-Nano-Mechanical Systems Cleanroom Laboratory. NCSA will assist with software development and design.

“We are very excited about the collaboration opportunities between nanoMFG Node efforts and NCSA researchers,” said Dr. Seid Koric, technical assistant director of NCSA and research professor of MechSE. “The advancement of cutting-edge tools for simulation software in hierarchical nanomanufacturing processes from nanoscale components to devices and systems is a huge accomplishment which will aid and enable a number of research projects at NCSA, the University of Illinois, and beyond.”

The NCSA Industry Program is the largest industrial HPC outreach effort in the world. The successful program operates dedicated non-government high-performance computing resources, including the iForge cluster.




The post U of Illinois, NCSA Launch First US Nanomanufacturing Node appeared first on HPCwire.

DOE Computational Science Graduate Fellows Begin Studies

Thu, 09/14/2017 - 16:27

Sept. 14 — A new class of 20 future HPC leaders is enrolled at U.S. universities this fall with support from the Department of Energy Computational Science Graduate Fellowship (DOE CSGF).

More than 425 students have participated in the program since it was established in 1991. Each is an advocate for computing’s capacity to advance science across a variety of disciplines.

Members of the fellowship’s twenty-seventh incoming class, their institutions and fields, are:

Peter Ahrens
Massachusetts Institute of Technology
Computer Science

Robert Baraldi
University of Washington
Applied Mathematics

Matthew Carbone
Columbia University
Chemical Physics

Gabriela Correa
Cornell University
Materials Science

Jennifer Coulter
Harvard University
Computational Materials Physics

Priya Donti
Carnegie Mellon University
Computer Science and Energy Policy

Annie Katsevich
New York University
Applied Mathematics

Jonas Kaufman
University of California, Santa Barbara
Computational Materials

Morgan Kelley
University of Texas
Process Systems

Claire Kopenhafer
Michigan State University

Alicia Magann
Princeton University

Quentarius Moore
Texas A&M University

Kari Norman
University of California, Berkeley

Miriam Rathbun
Massachusetts Institute of Technology
Computational Reactor Physics

Kevin Silmore
Massachusetts Institute of Technology
Chemical Engineering

Benjamin Toms
Colorado State University
Atmospheric Science

Steven Torrisi
Harvard University
Materials Physics

Annie Wei
Massachusetts Institute of Technology
Quantum Information/Quantum Algorithms

Zachary Weiner
University of Illinois at Urbana-Champaign
Cosmology, High Energy

Malia Wenny
Harvard University

Fellows receive a yearly stipend; full payment of university tuition and required fees (during the appointment period); and an annual academic allowance. Renewable for up to four years, the fellowship is guided by a comprehensive program of study that requires coursework in a science or engineering discipline plus computer science and applied mathematics. It also includes a three-month research practicum at one of 21 Department of Energy laboratories or sites across the country.

Additional details for each fellow are available via the program’s online fellow directory (https://www.krellinst.org/csgf/fellows/listing-by-program-year). For further information, contact the Krell Institute, DOE CSGF program manager, at https://www.krellinst.org/csgf/contact-us.

Source: DOE Computational Science Graduate Fellowship

The post DOE Computational Science Graduate Fellows Begin Studies appeared first on HPCwire.

PGI Rolls Out Support for Volta 100 in its 2017 Compilers and Tools Suite

Thu, 09/14/2017 - 13:00

PGI today announced a fairly lengthy list of new features to version 17.7 of its 2017 Compilers and Tools. The centerpiece of the additions is support for the Tesla Volta 100 GPU, Nvidia’s newest flagship silicon announced in April and now shipping to customers.

The Volta V100, developed at a cost of $3 billion, is a giant chip, 33 percent larger than the Pascal P100 and once again “the biggest GPU ever made.” Fabricated by TSMC on a custom 12-nm FFN high performance manufacturing process, the V100 GPU squeezes 21.1 billion transistors and almost 100 billion via connectors on an 815 mm2 die, about the size of the Apple watch,  (See HPCwire articles, Nvidia’s Mammoth Volta GPU Aims High for AI, HPC and First Volta-based Nvidia DGX Systems Ship to Boston-based Healthcare Providers)

Overall, says PGI, the new features in version 17.7 will help deliver improved performance and programming simplicity to high-performance computing (HPC) developers who target multicore CPUs and heterogeneous GPU-accelerated systems.

“We’re seeing a 1.5-2x performance improvement in OpenACC programs compiled with PGI 17.7 on Volta compared to Pascal,” said Michael Wolfe of Nvidia’s PGI compilers & tools group. “The support in PGI 17.7 for OpenACC with CUDA Unified Memory simplifies initial porting of applications to GPUs, the improved data handling for C++14 lambdas in OpenACC is really important to many C++ programmers, and the PGI Unified Binary for OpenACC to target both multicore CPUs and GPUs is a great new feature for ISVs that need to deliver a single GPU-accelerated binary to all of their customers.”

In this latest release, PGI OpenACC and CUDA Fortran now support Volta GV100 GPU, “offering more memory bandwidth, more streaming multiprocessors, next-generation Nvidia NVLink and new microarchitectural features that add up to better performance and programmability” according to PGI which is owned by Nvidia.

PGI 17.7 compilers also now leverage CUDA Unified Memory to simplify OpenACC programming on GPU-accelerated systems. When OpenACC allocatable data is placed in CUDA Unified Memory using a simple compiler option, no explicit data movement code or directives are needed.

Other key new features of the PGI 17.7 Compilers & Tools include:

  • OpenMP 4.5 for Multicore CPUs– Initial support for OpenMP 4.5 syntax and features allows the compilation of most OpenMP 4.5 programs for parallel execution across all the cores of a multicore CPU system. TARGET regions are implemented with default support for the multicore host as the target, and PARALLEL and DISTRIBUTE loops are parallelized across all OpenMP threads.
  • Automatic Deep Copy of Fortran Derived Types– Movement of aggregate, or deeply nested Fortran data objects between CPU host and GPU device memory, including traversal and management of pointer-based objects, is now supported using OpenACC directives.
  • C++ Enhancements– The PGI 17.7 C++ compiler includes incremental C++17 features, and is supported as a CUDA 9.0 NVCC host compiler. It delivers an average 20 percent performance improvement on the LCALS loops benchmarks.
  • Use C++14 Lambdas with Capture in OpenACC Regions–  C++ lambda expressions provide a convenient way to define anonymous function objects at the location where they are invoked or passed as arguments. Starting with the PGI 17.7 release, lambdas are supported in OpenACC compute regions in C++ programs, for example to drive code generation customized to different programming models or platforms.  C++14 opens doors for more lambda use cases, especially for polymorphic lambdas. Those capabilities are now usable in OpenACC programs.
  • Interoperability with the cuSOLVER Library– call optimized cuSolverDN routines from CUDA Fortran and OpenACC Fortran, C and C++ using the PGI-supplied interface module and the PGI-compiled version of the cuSOLVER library bundled with PGI 17.7.
  • PGI Unified Binary for NVIDIA Tesla and Multicore CPUs– use OpenACC to build applications for both GPU acceleration and parallel execution on multicore CPUs. When run on a GPU-enabled system, OpenACC regions offload and execute on the GPU. When run on a system without GPUs installed, OpenACC regions execute in parallel across all CPU cores in the system.
  • New Profiling features for CUDA Unified Memory and OpenACC– The PGI 17.7 Profiler adds new OpenACC profiling features including support on multicore CPUs with or without attached GPUs, and a new summary view that shows time spent in each OpenACC construct. New CUDA Unified Memory features include correlating CPU page faults with the source code lines where the associated data was allocated, support for new CUDA Unified Memory page thrashing, throttling and remote map events, NVLink support and more.

Other features and enhancements of PGI 17.7 include comprehensive support for environment modules on all supported platforms, prebuilt versions of popular open source libraries and applications, and new “Introduction to Parallel Computing with OpenACC” video tutorial series. PGI 17.7 is available for download today from the PGI website to all PGI Professional customers with active maintenance.

PGI includes high-performance parallel Fortran, C and C++ compilers and tools for x86-64 and OpenPOWER CPU processor-based systems and NVIDIA Tesla GPU Accelerators running Linux, Microsoft Windows or Apple macOS operating systems.

Link to complete list of PGI 17.7 features: http://www.pgicompilers.com/products/new-in-pgi.htm

The post PGI Rolls Out Support for Volta 100 in its 2017 Compilers and Tools Suite appeared first on HPCwire.

DARPA Pledges Another $300 Million for Post-Moore’s Readiness

Thu, 09/14/2017 - 11:24

Yesterday, the Defense Advanced Research Projects Agency (DARPA) launched a giant funding effort to ensure the United States can sustain the pace of electronic innovation vital to both a flourishing economy and a secure military. Under the banner of the Electronics Resurgence Initiative (ERI), some $500-$800 million will be invested in post-Moore’s law technologies that will benefit military and commercial users and contribute crucially to national security in the 2025 to 2030 time frame.

First made public in June (see HPCwire coverage here), ERI took shape over the summer as DARPA’s Microsystems Technology Office sought community involvement on the path forward for future progress in electronics. Based on that input, DARPA developed six new programs which are part of the overall larger vision of the Electronic Resurgence Initiative. The six programs are detailed in three Broad Agency Announcements (BAAs) published yesterday on FedBizOpps.gov. Each of the BAAs correlates to one of the ERI research pillars: materials and integration, circuit design, and systems architecture.

Planned investment is in the range of $200 million a year over four years. “ERI Page 3 Investments” refers to research areas that Gordon Moore predicted would become important for future microelectronics progress, cited on page 3 of Moore’s famous 1965 paper, “Cramming More Components onto Integrated Circuits.”

Also joining the ERI portfolio are several existing DARPA programs (including HIVE and CHIPS) as well as the Joint University Microelectronics Program (JUMP), a research effort in basic electronics education co-funded by DARPA and Semiconductor Research Corporation (SRC), an industry consortium based in Durham, N.C.

DARPA says that with the official roll out of the Electronics Resurgence Initiative, it “hopes to open new innovation pathways to address impending engineering and economics challenges that, if left unanswered, could challenge what has been a relentless half-century run of progress in microelectronics technology.”

DARPA is of course referring to the remarkable engine of innovation that is Moore’s law. Gordon Moore’s 1965 observation that transistor densities were doubling at roughly 24-month intervals set the stage for five decades of faster and cheaper microelectronics. But as node feature sizes approach the fundamental limits of physics, the design work and fabrication becomes ever more complex and expensive, jeopardizing the economic benefits of Moore’s dictum.

It’s something of a grand experiment, explained Bill Chappell, director of the Agency’s Microsystems Technology Office (MTO) in a press call, referring to the scale and scope of the Electronics Resurgence Initiative. DARPA has packaged up into one large announcement six different programs (released in three Broad Agency Announcements – BAAs — on FBO.gov). The six different programs will in sum receive $75 million in investment over the next year alone and on the order of $300 million over four years. Like all DARPA programs, the longevity and funding levels of these programs will be tied to performance.

“If we see that we’re getting broad resonance within the commercial industry and within the DoD industry, and unique partnerships are forming and/or unique capabilities are popping up for national defense, it will continue with the expectation or even grow,” said Chappell.

The DoD is finding it increasingly difficult to manufacture and design circuits, partly due to Moore’s law slowdowns and partly due to the scale of designs. “We are victim of our own success in that we have so many transistors available that we now have another problem which is complexity, complexity of manufacturing and complexity of design,” said Chappell. “So whether Moore’s law ends or not, at the DoD, from a niche development perspective we already have a problem on our hands. And we’re sharing that with the commercial world as well; you see a lot of mergers and acquisitions and tumult in the industry as they try to also grapple with some of the similar problems and the manpower required to get a design from concept into a physical product.”

Here’s a rundown on the six programs organized by their research thrust:

Materials and Integration (link)

  • Three Dimensional Monolithic System-on-a-Chip (3DSoC): Develop 3D monolithic technology that will enable > 50X improvement in SoC digital performance at power.
  • Foundations Required for Novel Compute (FRANC): Develop the foundations for assessing and establishing the proof of principle for beyond von Neumann compute topologies enabled by new materials and integration.

Design (link)

  • Intelligent Design of Electronic Assets (IDEA): “No human in the loop” 24-hour layout generation for mixed signal integrated circuits, systems-in-package, and printed circuit boards.
  • Posh Open Source Hardware (POSH): An open source System on Chip (SoC) design and verification eco-system that enables cost effective design of ultra-complex SoCs.

Novel Computing Architectures (link)

  • Software Defined Hardware (SDH): Build runtime reconfigurable hardware and software that enables near ASIC performance without sacrificing programmability for data-intensive algorithms.
  • Domain-Specific System on Chip (DSSoC): Enable rapid development of multiapplication systems through a single programmable device.

Chappell gave additional context for the Software Defined Hardware program, noting that it will look at course-grained reprogrammability specifically for big data programs. “We have the TPU and the GPU for dense problems, for dense searches, and dense matrix manipulation. We have recently started the HIVE program, which does sparse graph search. But the big question that still exists is what if you have a dense and sparse dataset? We don’t have a chip under development or even concepts that are very good at doing both of those types of datasets.”

What DARPA is envisioning is a reprogrammable system, or chip, that is intelligent enough and has an intelligent enough just in time compiler to recognize the data and type of data it needs to operate on and reconfigure itself to the need of that moment. DARPA has done seedlings to demonstrate that it’s feasible but “it’s still a DARPA-hard concept to pull off,” said Chappell.

DARPA will hold a number of Proposers Days to meet with interested researchers. The FRANC program of the Materials and Integration thrust will be run in the form of a webinar on Sept.15 and that thrust’s other program, 3DsoC, will take place at DARPA headquarters in Arlington, Va., on Sept. 22. The Proposers Day for the Architectures thrust’s two programs, DSSoC and SDH, will take place near DARPA headquarters in Arlington, Va., on Sept. 18 and 19, respectively. The Proposers Days for both programs in the Design thrust—IDEA and POSH—will take place on Sept. 22, in Mountain View, Calif. Details about all of these Proposers Day events and how to register are included in this Special Notice, DARPA-SN-17-75, posted on FBO.gov.

Asked about the goals for ERI writ large, Chappell said, “Overall success will look like we’ve invented the ideas that will be part of that 2025 and 2030 electronics community in such a way that both our defense base has better access to technology, better access to IP, better design services and capabilities than they have today because of these relationships that we are trying to build while simultaneously US interests in electronics in regards to economic development, maintaining our dominant global position is secured because of the new ideas that we are creating through these investments.

“These $75 million next year and $300 million over the course of the next four years that we’re planning is for very far-out research which often times is not something that a commercial entity can do because of its speculative nature and/or not something the DoD can do because it isn’t necessarily solving a today problem, but a tomorrow problem.”

DARPA is known for funding high-risk, high-reward R&D with broad commercial impact, helping to invent both the Internet and GPS.

The post DARPA Pledges Another $300 Million for Post-Moore’s Readiness appeared first on HPCwire.

IBM Breaks Ground for Complex Quantum Chemistry

Thu, 09/14/2017 - 11:10

IBM yesterday reported in Nature Communications the use of a novel algorithm to simulate BeH2 (beryllium-hydride) on a quantum computer. This is the largest molecule so far simulated on a quantum computer. The technique, which used six qubits of a seven-qubit system, is an important step forward and may suggest an approach to simulating ever larger molecules.

“Instead of forcing previously known classical computing methods onto quantum hardware, the scientists reversed the approach by building an algorithm suited to the capability of the current available quantum devices. This allows for extracting the maximal quantum computational power to solve problems that grow exponentially more difficult for classical computers,” according to the IBM announcement.

Quantum chemistry has long been regarded as of the great promises of quantum computing. A good example is nitrogen fixation – essentially making ammonia – from the air. Bacteria do it effortlessly. Industry still does it with a hundred-year-old Haber process, which is used mostly in fertilizer production today.

Today, simulating even small molecules with the needed accuracy to predict energy states and reactivity is hard. IBM performed the numerical simulation on H2, LiH, and BeH2. “While this model of BeH2 can be simulated on a classical computer, IBM’s approach has the potential to scale towards investigating larger molecules that would traditionally be seen to be beyond the scope of classical computational methods, as more powerful quantum systems get built,” noted IBM.

Here’s a good statement of the problem and IBM’s solution taken from the paper (Hardware-efficient variational quantum eigensolver for small molecules and quantum magnets):

“Finding exact solutions to such problems numerically has a computational cost that scales exponentially with the size of the system, and Monte Carlo methods are unsuitable owing to the fermionic sign problem. These limitations of classical computational methods have made solving even few-atom electronic-structure problems interesting for implementation using medium-sized quantum computers. Yet experimental implementations have so far been restricted to molecules involving only hydrogen and helium.

“Here we demonstrate the experimental optimization of Hamiltonian problems with up to six qubits and more than one hundred Pauli terms, determining the ground-state energy for molecules of increasing size, up to BeH2. We achieve this result by using a variational quantum eigenvalue solver (eigensolver) with efficiently prepared trial states that are tailored specifically to the interactions that are available in our quantum processor, combined with a compact encoding of fermionic Hamiltonians and a robust stochastic optimization routine.”

There are, of course many approaches to quantum computing and new kinds of qubits seem to appear weekly. IBM, Microsoft, and Google are focused on so-called universal quantum computers able to do pretty much anything classical computers can do. D-Wave builds quantum annealing computers best suited for certain optimization problems, some of which include quantum chemistry problems.

“IBM, Google and a number of academic labs have chosen relatively mature hardware, such as loops of superconducting wire, to make quantum bits (qubits). These are the building blocks of a quantum computer: they power its speedy calculations thanks to their ability to be in a mixture (or superposition) of ‘on’ and ‘off’ states at the same time.”[i] Microsoft is pursing one of the more exotic approaches – a topological qubit, the Majorana, a particle whose existence has been debated but for which evidence has been rapidly accumulating recently.

As described by IBM’s work, the fundamental goal in electronic-structure problems is to solve for the ground-state energy of many-body interacting fermionic Hamiltonians. Solving this problem on a quantum computer relies on a mapping between fermionic and qubit operators, which restates the problem as a specific instance of a local Hamiltonian problem on a set of qubits.

“Here we introduce and implement a hardware-efficient ansatz preparation for a VQE (variational quantum eigensolvers), whereby trial states are parameterized by quantum gates that are tailored to the physical device that is available. We show numerically the viability of such trial states for small electronic-structure problems and use a superconducting quantum processor to perform optimizations of the molecular energies of H2, LiH and BeH2, and extend its application to a Heisenberg antiferromagnetic model in an external magnetic field,” write the authors, all from IBM Research.[ii]

Below is diagram and caption of the recent work taken from the paper. (inset after)


Figure 1 | Quantum chemistry on a superconducting quantum processor. Solving electronic-structure problems on a quantum computer relies on mappings between fermionic and qubit operators. a, Parity mapping of eight spin orbitals (drawn in blue and red, not to scale) onto eight qubits, which are then reduced to six qubits owing to fermionic
spin and parity symmetries. The length of the bars indicate the parity of the spin orbitals that are encoded in each qubit. b, False-coloured optical micrograph of the superconducting quantum processor with seven transmon qubits. These qubits are coupled via two coplanar waveguide resonators (violet) and have individual coplanar waveguide resonators for control and read-out. c, Hardware-efficient quantum circuit for trial- state preparation and energy estimation, shown here for six qubits. For each iteration k, the circuit is composed of a sequence of interleaved single-qubit rotations Uq,d(θk) and entangling unitary operations UENT that entangle all of the qubits in the circuit. A final set of post-rotations
(I, X−p/2 or Yp/2) before the qubits are read out is used to measure the expectation values of the individual Pauli terms in the Hamiltonian and to estimate the energy of the trial state. d, An example of the pulse sequence for the preparation of a six-qubit trial state, in which UENT is implemented as a sequence of two-qubit cross-resonance gates.

IBM has certainly been an industry leader in providing access to quantum computing, most visibly through its IBM Q initiative  launched a year ago with a robust five-qubit quantum computer on the cloud for anyone to freely access; it has recently upgraded to a 16-qubit processor available for beta access.

To help showcase how quantum computers are adept to simulating molecules, developers and users of the IBM Q experience are now able to access a quantum chemistry Jupyter Notebook. The open source quantum chemistry Jupyter Notebook (available through the open access QISKit github repo) allows users to explore a method of ground state energy simulation for small molecules such as hydrogen and lithium hydride.

Quoted in the IBM announcement of the most recent work, Alán Aspuru-Guzik, professor of chemistry and chemical biology at Harvard University characterized IBM’s recent work as impressive, noting “When quantum computers are able to carry out chemical simulations in a numerically exact way, most likely when we have error correction in place and a large number of logical qubits, the field will be disrupted. Exact predictions will result in molecular design that does not need calibration with experiment. This may lead to the discovery of new small-molecule drugs or organic materials.”

Link to IBM paper: http://www.nature.com/nature/journal/v549/n7671/full/nature23879.html?foxtrotcallback=true

[i] http://www.nature.com/news/inside-microsoft-s-quest-for-a-topological-quantum-computer-1.20774

[ii] Abhinav Kandala, Antonio Mezzacapo, Kristan temme, Maika takita, Markus Brink, Jerry M. Chow1 & Jay M. Gambetta


The post IBM Breaks Ground for Complex Quantum Chemistry appeared first on HPCwire.

Scientists Using Intel-Cray ‘Theta’ Supercomputer to Map Brain Function

Thu, 09/14/2017 - 07:10

Sept. 11, 2017 — A neuroscientist and a computational scientist walk into a synchrotron facility to study a mouse brain… Sounds like a great set-up for a comedy bit, but there is no punchline. The result is cutting-edge science that can only be accomplished in a facility as scientifically integrated as the U.S. Department of Energy’s (DOE) Argonne National Laboratory.

At a casual, or even a more attentive glance, Doga Gursoy and Bobby Kasthuri would seem at opposite ends of the research spectrum. Gursoy is an assistant computational scientist at Argonne’s Advanced Photon Source (APS), a DOE Office of Science User Facility; Kasthuri, an Argonne neuroscientist.

But together, they are using Argonne’s vast arsenal of innovative technologies to map the intricacies of brain function at the deepest levels, and describing them in greater detail than ever before through advanced data analysis techniques.

Gursoy and Kasthuri are among the first group of researchers to access Theta, the new 9.65 petaflops Intel-Cray supercomputer housed at the Argonne Leadership Computing Facility (ALCF), also a DOE Office of Science User Facility. Theta’s advanced and flexible software platform supports the ALCF Data Science Program (ADSP), a new initiative targeted at big data problems, like Gursoy and Kasthuri’s brain connectome project.

ADSP projects explore and improve a variety of computational methods that will enable data-driven discoveries across all scientific disciplines.

“By developing and demonstrating rapid analysis techniques, such as data mining, graph analytics and machine learning, together with workflows that will facilitate productive usage on our systems for applications, we will pave the way for more and more science communities to use supercomputers for their big data challenges in the future,” said Venkat Vishwanath, ALCF Data Sciences Group Lead.

All about the connections

This new ADSP study of connectomes maps the connections of every neuron in the brain, whether human or mouse. Determining the location of every cell in the brain and how they communicate with each other is a daunting task, as each cell makes thousands of connections. The human brain, for example, has some 100 billion neurons, creating 100 trillion connections. Even the average mouse brain has 75 million neurons.

This ALCF award targets big data problems and our application of brain imaging does just that,” said Gursoy, assistant computational scientist in the X-Ray Science Division of Argonne’s Advanced Photon Source. “The basic goal is simple — we would like to be able to image all of the neurons in the brain — but the datasets from X-rays and electron microscopes are extremely large. They are at the tera- and petabyte scales. So we would like to use Theta to build the software and codebase infrastructure in order to analyze that data.”

This research was supported by the U.S. Department of Energy’s Office of Science. A portion of the work was also supported by Argonne’s Laboratory-Directed Research and Development (LDRD) program.

The process begins with two imaging techniques that will provide the massive sets of data for analysis by Theta. One is at the APS, where full brains can be analyzed at submicron resolution — in this case, the brain of a petite shrewmouse — through X-ray microtomography, a high-resolution 3-D imaging technique. Argonne’s X-ray Sciences Division of the APS provides the expertise for the microtomography research. Much like a CT scanner, it produces images as micro-thin slices of a material whose structure can be meticulously scrutinized. While this resolution provides a detailed picture of blood vessels and cell bodies, the researchers aim to go still deeper.

That depth of detail requires the use of an electron microscope, which transmits a short-wavelength electron beam to deliver resolution at the nanometer scale. This will allow for the capture of all the synaptic connections between individual neurons at small targeted regions guided by the X-ray microtomography.

For years, scientists at the APS have used these techniques to deepen our understanding of a wide variety of materials, from soil samples to new materials to biological matter,” said Kamel Fezzaa from sector 32-ID at the APS. “By coordinating our efforts with Argonne high-speed computing capabilities for this project, we are able to provide some truly revolutionary images that could provide details about brain functions that we have never before been able to observe.”

Both techniques can produce petabytes of information a day and, according to the researchers, the next generations of both microscopes will increase that amount dramatically.

Images produced by these datasets have to be processed, reconstructed and analyzed. Through the ADSP, Gursoy and Kasthuri are developing a series of large-scale data and computational steps — a pipeline — that integrates exascale computational approaches into an entirely new set of tools for brain research.

Taming of the shrew

The first case study for this pipeline is the reconstruction of an entire adult shrewmouse brain, which, they estimate, will produce one exabyte of data, or one billion gigabytes. And the studies only get bigger from there.

Machine learning will go through these datasets and help come up with predictive models. For this project, it can help with segmentation or reconstruction of the brain and help classify or identify features of interest,” said Vishwanath.

Lessons learned from the smaller shrewmouse brain will be applied to a large mouse brain, which constitutes a 10-fold increase in volume. Comparisons between the two will reveal how organizational structures form during development, from embryo to adult, and how they evolve. The reconstruction of a non-human primate brain, with a volume 100 times larger than a mouse brain, is being considered for a later study.

A neuroscientist and a computational scientist leave a synchrotron facility with studies from a mouse brain . . .  armed with new techniques to analyze this data. The images produced by their work will provide a clearer understanding of how even the smallest changes to the brain play a role in the onset and evolution of neurological diseases, such as Alzheimer’s and autism, and perhaps lead to improved treatments or even a cure.

Argonne National Laboratory seeks solutions to pressing national problems in science and technology. The nation’s first national laboratory, Argonne conducts leading-edge basic and applied scientific research in virtually every scientific discipline. Argonne researchers work closely with researchers from hundreds of companies, universities, and federal, state and municipal agencies to help them solve their specific problems, advance America’s scientific leadership and prepare the nation for a better future. With employees from more than 60 nations, Argonne is managed by UChicago Argonne, LLC for the U.S. Department of Energy’s Office of Science.

The U.S. Department of Energy’s Office of Science is the single largest supporter of basic research in the physical sciences in the United States and is working to address some of the most pressing challenges of our time. For more information, visit the Office of Science website.

Source: John Spizzirri, Argonne National Laboratory

The post Scientists Using Intel-Cray ‘Theta’ Supercomputer to Map Brain Function appeared first on HPCwire.

Lustre Enhances Flexibility for Big Data Era

Wed, 09/13/2017 - 16:51

Researchers at Oak Ridge National Laboratory (ORNL) and Intel Corporation have wrapped up a three year project aimed at giving users of Lustre, the Department of Energy’s preferred parallel file system, more flexibility. And the results are impressive.

Lustre is the preferred file system for the leadership scientific computing community for a simple reason: it has an unprecedented ability to store and retrieve the large-scale data inherent in complex scientific simulations such as those run at ORNL’s Leadership Computing Facility, home to Titan, the nation’s most powerful system for open science.

In the era of big data, however, there is no such thing as too much flexibility, and to bridge the gap between high-performance computing and today’s massive datasets, the ORNL/Intel team sought to modify the underlying Lustre code so that DOE’s trademark file system could better accommodate the data analytics workloads playing an increasingly important role in scientific discovery.

Their solution: Progressive File Layout (PFL), a novel storage scheme that absolves Lustre users of the responsibility of striping, or the method by which data is divided and stored across servers. The PFL approach gives users more opportunities to take advantage of Lustre’s highly scalable input/output performance, especially for big data-type workloads, and evolves Lustre to more easily facilitate large-scale datasets.

Specifically, PFL allows files to be striped dynamically depending on their size, at which point another striping scheme is implemented as the file size grows, and so on as the files surpass various size thresholds. “The scheme manages capacity based on layout and removes a significant responsibility from users,” said OLCF File Systems Team Lead Sarp Oral.

PFL is particularly relevant in the age of big data, as analytic and machine learning algorithms are increasingly read-heavy; i.e., they repeatedly read the same datasets, a process that creates “hot spots” in the form of requests repeatedly hitting the same resource. PFL paves the way for future Lustre enhancements to spread these requests out by creating replicas for the multiple requests, a process known as “file level replication,” therefore eliminating hot spots and improving the overall application I/O performance.

It’s a powerful modification—by giving users better flexibility on how to lay out their files, they can by extension take better advantage of Lustre’s unique capabilities without becoming parallel file system experts. And while Lustre has historically looked to underlying storage hardware for reliability, PFL also enables future developments aimed at providing reliability within the file system itself, allowing for the development of more scalable, efficient Lustre systems in less time with less money.

“While PFL is valuable on its own, it also enables future technology development by allowing Lustre to become an enterprise file system, expanding its use cases and marketability,” said Oral.

Large-scale testing of PFL was performed on Titan in June and was “very successful, with the new code showing significant improvement in file I/O performance,” said Neena Imam, deputy director of Collaborations for ORNL’s Computing and Computational Sciences Directorate. She added that despite its relative youth, “a stable PFL version is now available as of Lustre version 2.10 and has received significant attention in the Lustre community, which will benefit greatly from this addition.”

PFL is the result of a three year effort in which ORNL co-defined the architecture with Intel, oversaw the development efforts, and performed extensive testing.

The post Lustre Enhances Flexibility for Big Data Era appeared first on HPCwire.

Panasas Appoints Rob Berman Vice President of Sales for the Americas

Wed, 09/13/2017 - 13:22

SUNNYVALE, Calif., Sept. 12, 2017 — Panasas, the leader in performance scale-out network-attached storage (NAS), today announced the appointment of Rob Berman as the company’s vice president of sales for the Americas. Berman will oversee all direct and VAR channel sales for the Americas and play a key role in driving sales strategies for accelerating growth.

Berman brings more than 30 years of relevant industry experience to this role, having held executive sales and marketing management roles at large enterprise organizations where he developed and implemented effective sales and marketing strategies.

“Rob is an excellent addition to the Panasas team; we look forward to tapping his sales and leadership expertise as we expand our market reach,” said Faye Pairman, CEO of Panasas. “In addition to our deployments in traditional and commercial high-performance computing markets, we’re seeing strong growth opportunities in media and entertainment, life sciences and manufacturing that require the use of high-performance storage systems. Rob’s ability to grow strategic customer accounts and partner relationships will be a key contribution to our success.”

Berman comes to Panasas from Teradata, where he served as vice president of marketing applications sales and services, leading the sales, services and consulting organization for the company’s Integrated Marketing Cloud across the Americas. Prior positions include vice president of global alliances at Teradata and industry vice president for the U.S. at NCR Corporation, where he led the U.S. data warehousing, advanced analytics and applications sales and services teams.

For more information about Panasas, visit www.panasas.com.

About Panasas

Panasas is the performance scale-out NAS leader for unstructured data, driving industry and research innovation by accelerating workflows and simplifying data management. Panasas ActiveStor appliances leverage the patented PanFS storage operating system and DirectFlow protocol to deliver performance and reliability at scale from an appliance that is as easy to manage as it is fast to deploy. Panasas storage is optimized for the most demanding workloads in life sciences, manufacturing, media and entertainment, energy, government as well as education environments, and has been deployed in more than 50 countries worldwide. For more information, visit www.panasas.com.

Source: Panasas

The post Panasas Appoints Rob Berman Vice President of Sales for the Americas appeared first on HPCwire.

Chelsio Demonstrates 100 Gigabit iSCSI Performance

Wed, 09/13/2017 - 13:21

SUNNYVALE, Calif., Sept. 13, 2017 — Chelsio Communications, Inc., a leading provider of high performance (10Gb/25Gb/40Gb/50Gb/100Gb) Ethernet adapters for storage networking, virtualized enterprise datacenters, cloud service installations, and cluster computing environments, today announced the ultra-performance iSCSI Ethernet storage offload capabilities of its line of T6 100GbE Unified Wire adapters at this year’s Storage Developer Conference.

The demonstration shows Chelsio T6 dual-port 100GbE iSCSI offload adapters deliver 100 Gigabits-per-second (Gbps) line-rate and 2.9 Million I/O Per Second (IOPS) at 4KB I/O size enabling iSCSI enterprise-class storage target solutions to deliver extreme performance built with volume, off-the-shelf hardware and software components. With half of the processing resources of iSCSI target systems available even at maximum performance, latency delta of only 15 μs between remote and local storage access, and a power rating of only 20W, Chelsio’s T6 adapters provides significant performance and efficiency gains to datacenters with power hungry applications.

Showcasing the iSCSI hardware offload capabilities of the Chelsio T6 Unified Wire adapters, the demonstration shows how all-flash and hybrid storage array OEMs can easily enable such arrays with highest-performance iSCSI target capability that is fully compatible with the existing iSCSI ecosystem and seamlessly supports the routability, scalability and reach advantages of iSCSI. The complete Chelsio T6 iSCSI benchmark paper detailing the hardware/software configuration used and results achieved is available here.

Chelsio T6 100G Unified Wire adapters enable enterprise storage systems that are purpose-built to deliver optimized storage performance for various application workloads in mission-critical virtualized, private cloud environments.  The extreme efficiency of T6-enabled storage systems lets IT departments balance ultra-performance, capacity and features with game-changing storage economics.  Chelsio-enabled storage systems also easily integrate into diverse enterprise IT environments by enabling SAN and NAS deployments from the same array through iSCSI and TCP/IP protocol offload for iSCSI block-level protocol, as well as NFS and CIFS file-level protocols.

“With the explosive growth in All-Flash storage arrays the network is becoming a bottleneck,” said Seamus Crehan, president of Crehan Research. “Chelsio’s T6 100G Unified Wire solution with iSCSI offload alleviates this bottleneck with both high-throughput and IOPS enabling high-performance low-latency access to Flash storage.”

“Our T6 Unified Wire adapters establish a groundbreaking performance bar for Ethernet storage by enabling a 100 Gbps line-rate and extreme IOPS performance for highly efficient iSCSI storage target solutions,” said Kianoosh Naghshineh, CEO at Chelsio Communications.  “T6 storage protocol offload technology allows 100 Gigabit Ethernet iSCSI to completely supersede proprietary storage fabrics, while enabling concurrent, future-proof deployment of new Ethernet storage protocols without forcing an infrastructure forklift upgrade.  With transition to All-Flash arrays, all storage systems are essentially HPC systems and require offload technology to realize the Flash performance capabilities and value.”

About Chelsio iSCSI

Chelsio’s Terminator 6 ASIC offers a high performance, robust fourth generation implementation of iSCSI protocol over 100G Ethernet.  T6 delivers a low latency, low CPU utilization, high bandwidth, and high IOPs implementation of the iSCSI protocol at 100Gb and beyond.  It is the only robust iSCSI implementation in the industry that has not changed for a decade, allowing the benefit of years of quality assurance and the only industry solution that has scaled to 100Gb and beyond, and thus is ideally suited for Flash applications.

Chelsio’s iSCSI implementation is in production with a variety of OEMs and support will be available from several distributions.  It is a feature rich solution that enables turning on of the iSCSI digest protocol with no penalties, use of standard Ethernet frames, and support for T10 protocol. Chelsio’s iSCSI is available either in full offload form using Chelsio’s iSCSI stack, in partial offload form using the OEM’s software stack, or support of all the standard iSCSI offerings in the common distributions.

About Chelsio Communications

Chelsio is a recognized leader in high performance (10Gb/25Gb/40Gb/50Gb/100Gb) Ethernet adapters for networking and storage within virtualized enterprise datacenters, public and private hyperscale clouds, and cluster computing environments.  With a clear emphasis on performance and delivering the only robust offload solution, as opposed to simple speeds and feeds, Chelsio has set itself apart from the competition.  The Chelsio Unified Wire fully offloads all protocol traffic, providing no-compromise performance with high packet processing capacity, sub-microsecond hardware latency and high bandwidth. Visit the company at www.chelsio.com, and follow the company on Twitter and Facebook.

Source: Chelsio

The post Chelsio Demonstrates 100 Gigabit iSCSI Performance appeared first on HPCwire.

CPC Makes Liquid Cooling More Secure

Wed, 09/13/2017 - 13:18

ST. PAUL, Minn., Sept. 12, 2017 – CPC (Colder Products Company), maker of quick disconnects (QDs) designed specifically for liquid cooling use, has enhanced its LQ Series with easier to connect, more robust non-spill connectors for high-performance computing and data center applications. 

“As HPC manufacturers and data center operators continue to migrate to liquid cooling for their equipment, we hear concerns about quick disconnects as a potential point of vulnerability in liquid cooling systems,” says Dennis Downs, business manager, liquid cooling. “In working with tech customers, CPC has developed quick disconnects specifically for liquid cooling applications. Our LQ Series couplings are leak free, durable enough to withstand both years of use and corrosive fluids, and easy to use in tight spaces.”

CPC’s well-known LQ2, LQ4 and LQ6 Series quick disconnects now all feature unique swivel configurations and an integrated thumb latch for easy, one-handed operation in confined spaces like the server racks of large data centers.

In addition to the swivel joints, CPC reduced the force-to-connect by more than 20 percent. The robust LQ4 and LQ6 QDs also can handle twice the original side load force and feature a 38% higher burst pressure rating than before, offering additional protection from rough handling or use.

All LQ Series quick disconnects also feature a multilobed seal for redundant protection against leakage and lasting shape retention during extended periods of connection. Multilobe seals provide greater sealing efficiency than standard O-rings while requiring less force to connect. The non-spill design allows disconnection under pressure without leaks—a critically important factor in protecting electronics from exposure to fluid and enabling hot swapping of equipment.

The upgraded LQ4 and LQ6 Series quick disconnects join the LQ2 Series, the highest-flow capacity 1/8-inch connector in the liquid cooling industry. LQ2 Series connectors offer a 22 percent better flow coefficient than other 1/8-inch connectors. Higher flow capacities reduce pressure drops by an average of 34 percent, optimizing liquid cooling system performance.

“Couplings and QDs borrowed from other industries can result in suboptimal performance,” said Downs. “Because the CPC LQ Series QDs were designed for liquid cooling specifically, they are compatible with liquid cooling solutions and resistant to valve corrosion. They’re also tested to 10,000 cycles—double that of some other connectors—so manufacturers and operators can use the LQ Series with confidence.”

For more information about the performance and versatility of the LQ Series or any of the other 10,000+ innovative connection solutions CPC offers, visit cpcworldwide.com.

Source: CPC

The post CPC Makes Liquid Cooling More Secure appeared first on HPCwire.

Cubes, Culture, and a New Challenge: Trish Damkroger Talks about Life at Intel—and Why HPC Matters More Than Ever

Wed, 09/13/2017 - 10:51

Trish Damkroger wasn’t looking to change jobs when she attended SC15 in Austin, Texas. Capping a 15-year career within Department of Energy (DOE) laboratories, she was acting Associate Director for Computation at Lawrence Livermore National Laboratory (LLNL). Her mission was to equip the lab’s scientists and research partners with resources that would advance their cutting-edge work in stockpile science and other critical fields.

But a chance conversation brought up the possibility of a career jump, and one thing led to another.

Trish Damkroger

Today, Damkroger is Vice President of Intel’s Data Center Group and General Manager of its Technical Computing Initiative. Her work helps shape Intel’s high-performance computing (HPC) products and services for the technical market segment. Under that umbrella are the next-generation platform technologies and frameworks that will take Intel toward exascale and advance the convergence of traditional HPC, big data, and artificial intelligence workloads.

Along with her new job, Damkroger and her husband have moved to Oregon and joined the state’s $3.35 billion wine and grape industry. He recently retired, and she commutes to work from their 12-acre winery on Bald Peak, 20 miles or so south of Intel’s Hillsboro facilities. The two met at an executive coaching program run by the University of California, Berkeley’s Haas School of Business.

Damkroger is also a certified coach and a strong advocate for women in science, technology, engineering, and math (STEM). She has played leadership roles in the industry’s annual Supercomputing Conference for more than a decade. She chaired SC14—the year of HPC Matters—as well as heading the SC15 Steering Committee, and leading the SC16 Diverse HPC Workforce Committee. She’s signed on as Vice Chair of SC18.

Trish, you’re very well known in the industry, but I wonder if you could tell us a bit about your background and career path. What were some of the steps that led you to where you are today?

My dad is an electrical engineer. I had an older brother and a younger brother, and when we were growing up, the expectation was always that we would go to college, we could go to any state school we wanted, and we could be any kind of engineer. Those were our choices.

Of course, I thought that was awful, but now, with my own kids, I sometimes think it would have been good to give them a little more clarity, a few more guard rails—one of them is in computer engineering, and the other is still figuring out his passion. In any case, I chose electrical engineering graduating from Cal Poly. My older brother is a computer engineer, and my younger brother is a mechanical engineer.

When I started out, I was fascinated by the Six Million Dollar Man and Bionic Woman television shows. I wanted to do robotics, and create prostheses that connected to the brain, and make that whole thing work. But I was a little before my time, and the programs to do that really weren’t there.

After graduation, I worked full time at Hewlett-Packard and got my Master’s at Stanford studying AI and neural networks, which were in their infancy. That’s always been a passion for me—to figure out how the brain and body work together and how we can make prosthetics that mimic real limbs. It’s cool to see that coming to fruition now.

So you had worked at HP?  

Yes. I left HP to marry my husband, who lived in Livermore, California and I took a job with Sandia National Laboratories. I worked at Sandia for 10 years, and left there in 2000 to manage a product line for an IT service management company.

After 9/11, I wanted to return to the national security sector. I missed the labs and the national security mission. Plus the company I worked for was relocating and I was not interested in moving.

So I went to Lawrence Livermore, and I loved it, and I never expected to leave. I had worked my way up the organizational ladder and was probably in the last position I would have at the laboratory—and I realized I didn’t want to do what I was doing for another 10 years.

I came to Intel because it’s a chance to do something totally different. I love new challenges. I love to learn new things, and I have more chances to do that at Intel. It’s a completely different mindset and a completely new skillset to learn. I feel like I could spend decades here and continue to learn and grow.

Intel is in the middle of everything. It’s just a tremendously exciting place to be.

How did you make the move to Intel? Were you recruited? Were you job hunting?

Not job hunting at all. I ran into Debra Goldfarb [formerly of Microsoft and now Intel’s Senior Director of Market Intelligence] at SC15, and Deb asked if I was attending a women’s recruiting event Intel was putting on. I was already booked, and wasn’t looking to change jobs, so I didn’t attend. I made one of those, “If the right job comes up, keep me in mind” comments, but I wasn’t that serious—it was more in a spirit of not wanting to close doors.

Well, Deb set up a dinner meeting for me with Diane Bryant [president of the Intel Data Center Group], and I loved Diane. I mean, who doesn’t love Diane? We connected. She pointed out that I was passionate when I talked about all the things I was doing outside my job, with women and STEM, with SC. But, she said, “I don’t hear that same passion when you talk about your work.”

She was right, and it was a real “Wow” moment. It made me aware and got me thinking.

Has the Intel culture surprised you in any ways? Is it different from what you expected?

I’ll share a story. At Livermore, I had a beautiful office and my own conference room. On my first day at Intel, they walked me to my office, which is a small cube, and I asked, “Is this temporary?” But Intel being very egalitarian, they said, “No, everyone has a cube. BK—CEO Brian Krzanich—has a cube.”

I knew Intel was very egalitarian, and I think it’s a good thing. I like that philosophy. It’s a part of the culture going back to [Bob] Noyce and [Gordon] Moore. But the cube was a surprise.

People warned me about the pace. I’ve always worked hard and long hours, so that hasn’t changed, but being in a worldwide company is different. I have lots of early morning and evening calls. Intel’s business is truly global, and it’s 24/7. You’re dealing with China, with Europe—you have to be available. I knew about it intellectually, but it’s different when you’re actually doing the 6 am and 8 pm calls.

Another thing I love about Intel, and it’s huge, is how open everyone has been. They’ve been very welcoming, very willing to throw me in the middle of everything very quickly, and have the confidence in me that I can represent Intel all over the world. I love it. It shows the trust they have in their people.

You’re an advocate for diversity in STEM, and I know Intel is out front on this issue.  Why is diversity so important?

The real importance of diversity in HPC is that we need more people to go into tech fields. Period. Demand is growing, and we can’t meet it with only white men. The other point is that we’re selling to a diverse market. If we’re not engineering for that diversity, we’re going to lose. Everybody loses.

I’m very supportive of women in STEM. I’m continuing to push that, and to coach women who are in male-dominated fields.

You’ve focused industry attention on why HPC matters. Could you talk a bit about why sustained federal investments in HPC are so crucial?

My one-sentence answer is that HPC is absolutely essential to national competitiveness. China recognizes this. China expects to be at exascale in 2020. They’re getting there first because they’re making the investments. They’re developing indigenous technologies, and seeing HPC as a core element of competitiveness.

HPC is important because it is the way we are going to solve problems in every field. If we want the US to be at the forefront of innovation, we have to continue to invest. If we aren’t making those upstream investments to drive HPC innovation, we will lose our competitive edge.

That’s manufacturing and financial modeling and drug discovery. It’s autonomous driving and cognitive computing and bulletproof cyber security. It’s curing cancer, managing the electrical grid and safeguarding the nuclear arsenal. It’s sustainable agriculture and precision medicine.

Our digital infrastructure is just as important as our highways and airports. We need all hands on deck to help the government’s policymakers and funders understand HPC’s importance and why we need to push forward. We have to expand the capacity to support the nation’s critical science and technology research—DOE systems are at greater than 90 percent capacity, and that’s hard to keep up because you have to bring the systems down for maintenance.

We need to educate funders and decision makers about the ways government investment funds the full ecosystem—the labs and universities to build the large machines, conduct the research, do the applied math for the models, develop the applications and algorithms, explore the new technologies, and do all the things that will be in everyday computing environments 5-10 years out, and in your smart phone and wearables after that. If we stop those investments, the middle of the pyramid eventually collapses and the innovation stops. That’s an outcome no one wants.

About the Author

Jan Rowell writes about technology trends and impacts in HPC, healthcare, life sciences, and other industries.

The post Cubes, Culture, and a New Challenge: Trish Damkroger Talks about Life at Intel—and Why HPC Matters More Than Ever appeared first on HPCwire.

Amazon Debuts New AMD-based GPU Instances for Graphics Acceleration

Tue, 09/12/2017 - 16:04

Last week Amazon Web Services (AWS) streaming service, AppStream 2.0, introduced a new GPU instance called Graphics Design intended to accelerate graphics. The new instance is based on AMD’s FirePro S7150x2 Server GPUs equipped with AMD Multiuser GPU technology. The AWS move is a win for AMD which has been on a roll of late with the launch of its EPYC chip line perhaps being the high point at least in HPC terms.

In making the announcement, AWS said the new instance type allows users to run graphics applications at a fraction of the cost of using graphics workstations, and can reduce the cost of streaming graphics applications with AppStream 2.0 by up to 50%. Achieving fast graphics performance in the cloud has long been challenging while boosting performance locally with high-end workstations is expensive.

“Graphics Design instances are ideal for delivering applications that rely on hardware acceleration of DirectX, OpenGL, or OpenCL, such as Adobe Premiere Pro, Autodesk Revit, and Siemens NX. With this launch, AppStream 2.0 now offers three graphics instance types – Graphics Design, Graphics Desktop, and Graphics Pro – optimized to support a broad selection of graphics workloads,” said AWS.

There are four Graphics Design instance sizes with different GPU and compute combinations that scale to support the performance requirements of a range of graphics applications such as engineering and creative design. The smallest instance size available “is large, with 2 vCPU, 7.5 GiB system memory, and 1 GiB graphics memory. The highest performing instance size available is 4xlarge with 16 vCPUs, 61 GiB system memory, and 8 GiB graphics memory. This range of instance sizes allows you to select the configuration that matches your application’s requirements and provide your users a fluid and high-performance experience.”

Writing in a blog today, Michael DeNeffe, AMD director of cloud graphics said, “I’m thrilled that AWS has chosen AMD Radeon Pro MxGPU (multiuser GPU) technology for their new Graphics Design instance type on Amazon AppStream 2.0.  Amazon AppStream 2.0 is a fully managed, secure application streaming service that allows you to stream desktop applications from AWS to any device running a web browser, without rewriting them. The new Graphics Design instance type runs on our S7150x2 GPU, the virtualized graphics workhorse of our Radeon Pro graphics product.”

Radeon Pro virtualized GPUs feature Multi-user GPU, which AMD says is the industry’s first and only hardware-based virtualization technology in a GPU, based on SR-IOV (Single Root I/O Virtualization). SR-IOV is a big deal for three reasons, says the company:

  • GPU scheduling logic in hardware helps to ensure better quality of service for users
  • It preserves the data integrity of Virtualized Machines (VM) and their application data through hardware-enforced memory isolation logic preventing one VM from being able to access another VM’s data
  • It exposes all graphics functionality of the GPU to applications allowing for full virtualization support for not only graphics APIs like DirectX and OpenGL but also GPU compute APIs like OpenCL.

It’s worth noting that AppStream 2.0 also has GPU instances based Nvidia GPUs (Graphics Desktop instance (K520) and Graphics Pro (Tesla M60)). AWS is the latest cloud computing company to use Radeon Pro technology, reports AMD, citing recent collaborations with Google and Alibaba.

Link to AWS announcement: https://aws.amazon.com/about-aws/whats-new/2017/09/introducing-amazon-appstream-2-graphics-design-a-new-lower-cost-instance-type-for-streaming-graphics-applications/

Link to AMD blog: https://pro.radeon.com/en-us/secure-mobility-amd-radeon-pro-gpus-bring-cutting-edge-graphics-virtualization-to-amazon-appstream-2-0/

Link to list of AppStream 2.0 instances: http://docs.aws.amazon.com/appstream2/latest/developerguide/instance-types.html

The post Amazon Debuts New AMD-based GPU Instances for Graphics Acceleration appeared first on HPCwire.

Chaminade University to Offer New Data Science Training for Hawaii Students

Tue, 09/12/2017 - 12:48

HONOLULU, Hawaii, Sept. 11, 2017 – Chaminade University is partnering with universities in Texas and Georgia on a two-year pilot program that trains Native Hawaiian and Pacific Islander (NHPI) students for careers in data science, analytics and visualization.

The SPICE (Supporting Pacific Indigenous Computing Excellence) project is funded by a $300,000 award from the National Science Foundation (NSF) through its INCLUDES (Inclusion across the Nation of Communities of Learners of Underrepresented Discoverers in Engineering and Science) initiative. INCLUDES is part of the NSF’s “10 Big Ideas” program.

“The vision of SPICE is training a cadre of students who will lead data science, visualization and analytics efforts that support health, sustainability and social justice in Hawaii and elsewhere in the Pacific,” said Dr. Helen Turner, the project’s co-principal investigator and Chaminade’s dean of Natural Sciences and Mathematics and Vice President for Innovation.

“Solutions to many critical regional problems lie in ‘big data,’” Dr. Turner said. “It’s key that Hawaii’s future science, technology and business leaders are prepared to use data science in their careers and advocacy.

“Analyzing and applying big data has the potential to change lives in Hawaii for the better,” Dr. Turner added, “and we want our students to be part of that better future.”

Partnering with Chaminade on SPICE are the Texas Advanced Computing Center (TACC) at The University of Texas at Austin and the Center for Education Integrating Science, Mathematics & Computing at the Georgia Institute of Technology. SPICE principal investigator is Kelly Gaither, TACC Director of Visualization.

Gaither explained that SPICE students will work with large data sets offering possible solutions to current and emerging problems in the Pacific, including health inequities, natural resource management and economic development. Moreover, preparing these students for data science and computational careers will support Hawaii’s transition to an innovation economy.

“The long-term goal is developing the SPICE partnership into a backbone organization that can frame the current and future efforts as an NSF INCLUDES Alliance,” according to Gaither, “starting with a one-month summer immersion program in 2018 and building to a data science curriculum at Chaminade.

Chaminade President Lynn Babington said one of the “most compelling” aspects of this data science, visualization and analytics  initiative is its broad applicability.

“These skills are needed by Hawaii’s future workforce across diverse sectors,” Dr. Babington said, “including business, science, health care and environmental protection. This is the gap Chaminade will address.”

Source: Chaminade University

The post Chaminade University to Offer New Data Science Training for Hawaii Students appeared first on HPCwire.

Google Announces 2017 PhD Fellowship Winners

Tue, 09/12/2017 - 11:50

Since 2009 Google has conducted a PhD Fellowship Program in support of various computational disciplines and upcoming talent enrolled in PhD programs. Today the web giant announced the 2017 Google Fellows who will work throughout the year and then attend next year’s Google Summit to discuss their work.

“Our PhD Fellows represent some the best and brightest young researchers around the globe in Computer Science and it is our ongoing goal to support them as they make their mark on the world,” wrote Susie Kim, Google Program Manager, University Relations, on Google’s research blog today. Google has so far supported more 300 graduate students in AustraliaChina and East AsiaIndiaNorth America, Europe and the Middle East “who seek to shape and influence the future of technology” according to Kim.

Under the terms of the awards, “Students receive named Fellowships which include a monetary award. The funds are given directly to the university to be distributed to cover the student’s expenses and stipend as appropriate. The funds are given as an unrestricted gift, and it is Google’s policy not to pay for overhead on unrestricted gifts. In addition, the student will be matched with a Google Research Mentor who we hope will become a valuable resource to the student. There is no employee relationship between the student and Google as a result of receiving the fellowship. Fellowship recipients are not subject to intellectual property restrictions unless they complete an internship at Google. Fellowship recipients serving an internship are subject to the same intellectual property and other contractual obligations as any other Google intern. If a Fellowship student is interested, an internship at Google is encouraged, but not guaranteed or required.”

Here is the 2017 Google PhD Fellows List:

Algorithms, Optimizations and Markets
Chiu Wai Sam Wong, University of California, Berkeley
Eric Balkanski, Harvard University
Haifeng Xu, University of Southern California

Human-Computer Interaction
Motahhare Eslami, University of Illinois, Urbana-Champaign
Sarah D’Angelo, Northwestern University
Sarah Mcroberts, University of Minnesota – Twin Cities
Sarah Webber, The University of Melbourne

Machine Learning
Aude Genevay, Fondation Sciences Mathématiques de Paris
Dustin Tran, Columbia University
Jamie Hayes, University College London
Jin-Hwa Kim, Seoul National University
Ling Luo, The University of Sydney
Martin Arjovsky, New York University
Sayak Ray Chowdhury, Indian Institute of Science
Song Zuo, Tsinghua University
Taco Cohen, University of Amsterdam
Yuhuai Wu, University of Toronto
Yunhe Wang, Peking University
Yunye Gong, Cornell University

Machine Perception, Speech Technology and Computer Vision
Avijit Dasgupta, International Institute of Information Technology – Hyderabad
Franziska Müller, Saarland University – Saarbrücken GSCS and Max Planck Institute for Informatics
George Trigeorgis, Imperial College London
Iro Armeni, Stanford University
Saining Xie, University of California, San Diego
Yu-Chuan Su, University of Texas, Austin

Mobile Computing
Sangeun Oh, Korea Advanced Institute of Science and Technology
Shuo Yang, Shanghai Jiao Tong University

Natural Language Processing
Bidisha Samanta, Indian Institute of Technology Kharagpur
Ekaterina Vylomova, The University of Melbourne
Jianpeng Cheng, The University of Edinburgh
Kevin Clark, Stanford University
Meng Zhang, Tsinghua University
Preksha Nama, Indian Institute of Technology Madras
Tim Rocktaschel, University College London

Privacy and Security
Romain Gay, ENS – École Normale Supérieure
Xi He, Duke University
Yupeng Zhang, University of Maryland, College Park

Programming Languages, Algorithms and Software Engineering
Christoffer Quist Adamsen, Aarhus University
Muhammad Ali Gulzar, University of California, Los Angeles
Oded Padon, Tel-Aviv University

Structured Data and Database Management
Amir Shaikhha, EPFL CS
Jingbo Shang, University of Illinois, Urbana-Champaign

Systems and Networking
Ahmed M. Said Mohamed Tawfik Issa, Georgia Institute of Technology
Khanh Nguyen, University of California, Irvine
Radhika Mittal, University of California, Berkeley
Ryan Beckett, Princeton University
Samaneh Movassaghi, Australian National University

Link to blog: https://research.googleblog.com

Link to Google Fellowship description: https://research.google.com/research-outreach.html#/research-outreach/graduate-fellowships

Feature image: attendees to this year’s Google Summit just completed. The 2017 awardees will attend next year’s conference.

The post Google Announces 2017 PhD Fellowship Winners appeared first on HPCwire.

TACC STEM Partnership Wins NSF INCLUDES Grant

Tue, 09/12/2017 - 11:04

AUSTIN, Sept. 12, 2017 — The Texas Advanced Computing Center (TACC) at The University of Texas at Austin today announced a $300,000 award for “Supporting Pacific Indigenous Computing Excellence (SPICE).” Funded by the National Science Foundation (NSF), Inclusion across the Nation of Communities of Learners of Underrepresented Discoverers in Engineering and Science (INCLUDES) initiative, the SPICE project will enhance U.S. leadership in science, technology, engineering and mathematics (STEM) discoveries and innovations by focusing on diversity, inclusion, and broadening participation in these fields.

“Broadening participation in STEM is necessary for the United States to retain its position as the world’s preeminent source of scientific innovation,” said NSF Director France Cordova.

“The National Science Foundation has a long history of working to address difficult challenges by creating the space for inventive solutions. NSF INCLUDES breaks new ground by providing a sustained commitment to collaborative change with the goal of bringing STEM opportunities to more people and more communities across the country,” Cordova said.

NSF INCLUDES is among the organization’s “10 Big Ideas for Future NSF Investments” research agendas that identify areas for future investment at the frontiers of science and engineering.

As part of the two-year INCLUDES SPICE pilot beginning September 1, 2017, and ending on August 30, 2019, TACC will partner with Chaminade University of Honolulu (CUH) and Georgia Institute of Technology’s Center for Education Integrating Science, Mathematics, & Computing (GIT-CEISMC).

The vision of “Supporting Pacific Indigenous Computing Excellence” is to train a cadre of high school and undergraduate students, particularly Native Hawai’ian and Pacific Islanders (NHPI), who will lead data science, visualization and analytics efforts that support health, sustainability and social justice in Hawai’i and the US-affiliated Pacific. Workforce development efforts in the state of Hawai’i have identified data science and computational careers as key to the state’s economic diversification and development going forward.

“Students in these communities are underrepresented in STEM fields, and specifically in cutting-edge data science, which limits self-determination,” said Kelly Gaither, principal investigator of the project and director of Visualization at TACC. “NHPI are broadly disenfranchised from STEM. In all areas of society, the digital divide is troubling, but for STEM education, it creates a new challenge.”

According to Gaither, these students will gain access to, and develop the ability to work with, large data sets, which will help provide solutions to current and emerging problems in the Pacific such as health inequities, natural resource management, and economic development.

“Agency over ‘big data’ sets that are relevant to Pacific issues, and contemporary skills in data science, analytics and visualization are the next frontier in educational disparities for this population of students,” Gaither said. “For example, students will learn how to conduct data analytics so they can do visual reasoning.”

The program will serve as a systematic effort to provide undergraduate preparation in data science, analytics and visualization in the few minority-serving, non-research intensive institutions whose missions address educational disparities.

“SPICE will establish a model for data science preparation that is contemporary, culturally-consistent and sustainable,” said Helen Turner, a professor of biology, vice president for Innovation at CUH, and chief education officer of the project.

“Students in the region face severe challenges in health, poverty, environmental resilience, and the erosion of traditional culture. These issues have been the focus of numerous programs to increase STEM participation, and they are of high importance,” Turner said.

SPICE is based on unique expertise and proven models established by TACC, CUH and GIT-CEISMC. TACC is an internationally recognized advanced computing, data science, and visualization center leading strategic diversity and outreach programs. CUH has undergone a striking institutional transformation in recent years with its mission-driven effort to become a leader in science education for indigenous populations in the Pacific. GIT-CEISMC has a world-class evaluation team with extensive experience with underserved populations.

“The long-term goal is to develop the TACC-CUH-GIT partnership into a backbone organization that can frame the current and future efforts as an NSF INCLUDES Alliance,” Gaither said.

SPICE has three goals: 1) perform original research and program development to bridge computation and culture; 2) implement a Data Science, Analytics and Visualization Summer Immersion Experience; and 3) build capacity in data science at a NHPI-serving undergraduate institution through the development of an undergraduate Certificate in Data Science.

The program will significantly improve curriculum, research and indigenous participation in data science in the Pacific region. For the Summer Immersion Experience, SPICE will train 25 high school students and undergraduates, drawing from the Federated State of Micronesia, Guam, American Samoa, Saipan, and the Marshall Islands. A new Certificate in Data Science will be developed, providing STEM majors at CUH the opportunity to be certified in these competency areas.

“These activities will begin to lessen the digital divide that threatens these communities at the structural level, while CUH and Pacific anthropology faculty will address the cultural divide that threatens NHPI participation in this emergent STEM field,” Turner said. “Through engaging community and family in all aspects of the project, we seek to address the alienation and outsider status that are often the lived experience of these students who choose STEM careers.”

The pilot phase will integrate the planned expansion to scale the SPICE program, with a strategy to engage all students in the region. SPICE will use advanced computing for social change relevant to these communities, bridging cultural and Western knowledge, and empowering indigenous self-determination.

According to Turner, curricula emphasizing the role of big data for ‘social good’ in information education settings have demonstrated success at engaging greater proportions of women and underrepresented minorities.

“Analyzing and applying big data has the potential to change lives in Hawai’i for the better,” Turner said, “and we want our students to be part of that better future.”

Source: TACC

The post TACC STEM Partnership Wins NSF INCLUDES Grant appeared first on HPCwire.

Los Alamos Laboratory Director Charles F. McMillan to Retire

Tue, 09/12/2017 - 11:02

LOS ALAMOS, N.M., Sept. 12, 2017 — Charles F. (Charlie) McMillan has informed employees of Los Alamos National Laboratory that he intends to step down as Laboratory Director at the end of this calendar year. The announcement was made by McMillan in person at an all-employee meeting earlier today at the Laboratory.

McMillan told employees, “It has truly been an honor and a privilege to serve as your Director these past six years. Every day, I have been in awe of the people of this great Laboratory and what we have been able to contribute to this nation’s security.”

“Charlie McMillan has led Los Alamos National Laboratory with a rare combination of commitment, intelligence and hard work,” said Norm Pattiz, Chairman of Los Alamos National Security, LLC (LANS). “Because of his passion for the Lab, its missions and its people, he agreed to stay on as Director at the Board’s request, past his originally planned retirement date. We appreciate Charlie’s commitment and believe he has put this iconic institution in a strong position to continue serving the country for many years to come.”

“Our work to advance analytical capabilities with the new Trinity supercomputer, our experimental capabilities at Los Alamos and at the National Nuclear Security Site in Nevada and our cutting-edge research in materials science have strengthened our nuclear weapons and global security mission work and paved the way for an enduring future for Los Alamos National Laboratory,” said McMillan.

“I am proud of the scientific and engineering work that underpins all of our national security efforts and leads to scientific advancements,” said McMillan. “From helping explore Mars, to aiding global efforts to develop an HIV vaccine, to producing life-saving medical isotopes, to earth system modeling, Los Alamos’ unique multidisciplinary scientific capabilities make the world a better place.”

McMillan noted in his all-employee meeting the health of the Laboratory, as measured both in hiring and budget. Los Alamos hired more than 1,000 employees last fiscal year (FY) and expects to hire roughly the same number by the end of FY2017. The Laboratory’s budget has grown approximately $400 million from FY2013 to today’s FY2017 budget of $2.5 billion.

“With an eye on the future, we have taken steps in the past few years to put Los Alamos in a strong position to meet the challenges that lie ahead,” said McMillan. “The work that has been done to develop future leaders at all levels, expand the operating budget, and hire the workforce of 2030 has put the Laboratory on the right trajectory for continued success in the decades to come.”

McMillan noted, “I am encouraged by the government’s commitment to ensuring the long-term viability of the nuclear weapons stockpile, and I am upbeat that funding levels for the Stockpile Stewardship Program are headed in the right direction. In addition, the long overdue modernization of the nuclear weapons complex and its infrastructure, including facilities at our Laboratory, now appears to be firmly underway.”

McMillan told staff that he would work closely with his successor, when named by the LANS Board, to ensure a smooth transition.

About Los Alamos National Laboratory

Los Alamos National Laboratory, a multidisciplinary research institution engaged in strategic science on behalf of national security, is operated by Los Alamos National Security, LLC, a team composed of Bechtel National, the University of California, BWXT Government Group, and URS, an AECOM company, for the Department of Energy’s National Nuclear Security Administration.

Los Alamos enhances national security by ensuring the safety and reliability of the U.S. nuclear stockpile, developing technologies to reduce threats from weapons of mass destruction, and solving problems related to energy, environment, infrastructure, health, and global security concerns.

Source: LANL

The post Los Alamos Laboratory Director Charles F. McMillan to Retire appeared first on HPCwire.

Cavium Joins OpenMP Effort

Tue, 09/12/2017 - 10:57

AUSTIN, Texas, Sept. 12, 2017 — The OpenMP ARB, a group of leading hardware and software vendors and research organizations that create the OpenMP standard parallel programming specification, today announced that Cavium, Inc., a leading provider of semiconductor products that enable intelligent processing for enterprise, data center, cloud, service provider wired and wireless networking has joined as a new member.

Cavium’s membership in the OpenMP ARB further highlights our strong belief in industry’s demand for parallel computing and the significance of the ARM Architecture,” said Avinash Sodani, Distinguished Engineer at Cavium. “Cavium’s strong product portfolio includes ThunderX2, a compelling server-class, multi-core ARMv8 CPU suited for the most demanding compute workloads. We look forward to working with other OpenMP members in furthering OpenMP standards to meet the challenges of the Exascale era.”

“We are pleased that Cavium, Inc. has joined the OpenMP ARB and will help strengthen OpenMP support in the ARM software ecosystem,” says Michael Klemm, CEO of the OpenMP ARB.

About OpenMP

The OpenMP ARB has a mission to standardize directive-based multi-language high-level parallelism that is performant, productive and portable. Jointly defined by a group of major computer hardware and software vendors, the OpenMP API is a portable, scalable model that gives parallel programmers a simple and flexible interface for developing parallel applications for platforms ranging from embedded systems and accelerator devices to multicore systems and shared-memory systems. The OpenMP ARB owns the OpenMP brand, oversees the OpenMP specification and produces and approves new versions of the specification. Further information can be found at http://www.openmp.org/.

About Cavium

Cavium, Inc. offers a broad portfolio of infrastructure solutions for compute, security, storage, switching, connectivity and baseband processing. Cavium’s highly integrated multi-core SoC products deliver software compatible solutions across low to high performance points enabling secure and intelligent functionality in Enterprise, Data center and Service Provider Equipment. Cavium processors and solutions are supported by an extensive ecosystem of operating systems, tools, application stacks, hardware reference designs and other products. Cavium is headquartered in San Jose, CA with design centers in California, Massachusetts, India, Israel, China and Taiwan. For more information, please visit http://www.cavium.com.

Source: OpenMP

The post Cavium Joins OpenMP Effort appeared first on HPCwire.

Sandia National Lab Researcher Bochev Wins CFD Award

Tue, 09/12/2017 - 10:57

Pavel Bochev, Sandia National Laboratories researcher, has been awarded the Thomas J.R. Hughes Medal by the U.S. Association for Computational Mechanics. The award is given biannually for “outstanding and sustained contributions to the broad field of computation fluid dynamics.”

CFD is of course crucial to many simulation applications. Bochev was chosen specifically for “foundational contributions to numerical partial differential equations, especially advances in the development and analysis of new stabilized and compatible finite element methods, and software design for advanced discretizations.” said Sandia manager Michael Parks

Pavel Bochev, Sandia

Finite element methods are a fundamental modeling and simulation tool for science and engineering problems. FEM’s applications range from examining the integrity of the nuclear stockpile to assessing the vulnerabilities and risks of natural and human systems to climate change.

FEM convert differential equations describing physical phenomena, such as fluid flows and electromagnetism, into algebraic equations that can be solved on a computer. Compatible FEM mimic the mathematical structure of these differential equations and can deliver robust and physically meaningful results. However, they also can be more difficult to implement and solve. Stabilized FEM use simpler, generic finite element definitions. But because approximations are involved, the design of successful elements requires careful analysis of the mathematical properties lost and the most appropriate mechanisms to recover them.

The award was presented at the 14th U.S. National Congress on Computational Mechanics in Montreal in mid-July.

As defined by the USACM, “The Thomas J. R. Hughes Medal is given in recognition of outstanding and sustained contributions to the broad field of Computational Fluid Dynamics (CFD). These contributions shall generally be in the form of important research results that significantly advance the understanding of theories and methods impacting CFD. Industrial applications and engineering analyses that advance CFD shall also represent accomplishments worthy of recognition. [This award replaces the former USACM Computational Fluid Mechanics Award.]”

Link to article on the Sandia web site: https://share-ng.sandia.gov/news/resources/news_releases/computational_award/#.WbgLWK2ZMlU

The post Sandia National Lab Researcher Bochev Wins CFD Award appeared first on HPCwire.

Julia Joins Petaflop Club

Tue, 09/12/2017 - 08:55

BERKELEY, Calif., Sept. 12, 2017 — Julia has joined the rarefied ranks of computing languages that have achieved peak performance exceeding one petaflop per second – the so-called ‘Petaflop Club.’

The Julia application that achieved this milestone is called Celeste.  It was developed by a team of astronomers, physicists, computer engineers and statisticians from UC Berkeley, Lawrence Berkeley National Laboratory, National Energy Research Scientific Computing Center (NERSC), Intel, Julia Computing and the Julia Lab at MIT.

Celeste uses the Sloan Digital Sky Survey (SDSS), a dataset of astronomical images from the Apache Point Observatory in New Mexico that includes every visible object from over 35% of the sky – hundreds of millions of stars and galaxies.  Light from the most distant of these galaxies has been traveling for billions of years and lets us see how the universe appeared in the distant past.

Since SDSS data collection began in 1998, the process of cataloging these stars and galaxies was painstaking and laborious.

So the Celeste team developed a new parallel computing method to process the entire SDSS dataset. Celeste is written entirely in Julia, and the Celeste team loaded an aggregate of 178 terabytes of image data to produce the most accurate catalog of 188 million astronomical objects in just 14.6 minutes with state-of-the-art point and uncertainty estimates.

Celeste achieved peak performance of 1.54 petaflops using 1.3 million threads on 9,300 Knights Landing (KNL) nodes of the Cori supercomputer at NERSC – a performance improvement of 1,000x in single-threaded execution.

The Celeste research team is already looking to new challenges. For example, the Large Synoptic Survey Telescope (LSST), scheduled to begin operation in 2019, is 14 times larger than the Apache Point telescope and will produce 15 terabytes of images every night. This means that every few days, the LSST will produce more visual data than the Apache Point telescope has produced in 20 years. With Julia and the Cori supercomputer, the Celeste team can analyze and catalog every object in those nightly images in as little as 5 minutes.

The Celeste team is also working to:

  • Further increase the precision of point and uncertainty estimates
  • Identify ever-fainter points of light near the detection limit
  • Improve the quality of native code for high performance computing

The Celeste project is a shining example of:

  • High performance computing applied to real-world problems
  • Cross-institutional collaboration including researchers from UC Berkeley, Lawrence Berkeley National Laboratory, National Energy Research Scientific Computing Center (NERSC), Intel, Julia Computing and the Julia Lab at MIT
  • Cross-departmental collaboration including astronomy, physics, computer science, engineering and mathematics
  • Julia, the fastest modern open source high performance programming language for scientific computing
  • Parallel and multithreading supercomputing capabilities
  • Public support for basic and applied scientific research

About Julia and Julia Computing

Julia is the fastest modern high performance open source computing language for data, analytics, algorithmic trading, machine learning and artificial intelligence. Julia combines the functionality and ease of use of Python, R, Matlab, SAS and Stata with the speed of C++ and Java. Julia delivers dramatic improvements in simplicity, speed, capacity and productivity. Julia provides parallel computing capabilities out of the box and unlimited scalability with minimal effort. With more than 1 million downloads and +161% annual growth, Julia is one of the top 10 programming languages developed on GitHub and adoption is growing rapidly in finance, insurance, energy, robotics, genomics, aerospace and many other fields.

Julia users, partners and employers hiring Julia programmers in 2017 include Amazon, Apple, BlackRock, Capital One, Comcast, Disney, Facebook, Ford, Google, Grindr, IBM, Intel, KPMG, Microsoft, NASA, Oracle, PwC, Raytheon and Uber.

  1. Julia is lightning fast. Julia provides speed improvements up to 1,000x for insurance model estimation, 225x for parallel supercomputing image analysis and 10x for macroeconomic modeling.
  2. Julia provides unlimited scalability. Julia applications can be deployed on large clusters with a click of a button and can run parallel and distributed computing quickly and easily on tens of thousands of nodes.
  3. Julia is easy to learn. Julia’s flexible syntax is familiar and comfortable for users of Python, R and Matlab.
  4. Julia integrates well with existing code and platforms. Users of C, C++, Python, R and other languages can easily integrate their existing code into Julia.
  5. Elegant code. Julia was built from the ground up for mathematical, scientific and statistical computing. It has advanced libraries that make programming simple and fast and dramatically reduce the number of lines of code required – in some cases, by 90% or more.
  6. Julia solves the two language problem. Because Julia combines the ease of use and familiar syntax of Python, R and Matlab with the speed of C, C++ or Java, programmers no longer need to estimate models in one language and reproduce them in a faster production language. This saves time and reduces error and cost.

Julia Computing was founded in 2015 by the creators of the open source Julia language to develop products and provide support for businesses and researchers who use Julia.

Source: Julia Computing

The post Julia Joins Petaflop Club appeared first on HPCwire.

NSF Announces 14 New PIRE Awards to Support Scientific Collaboration in 24 Countries

Tue, 09/12/2017 - 08:53

Sept. 12, 2017 — The National Science Foundation (NSF) is pleased to announce 14 new Partnerships for International Research and Education (PIRE) awards, totaling more than $66 million over the next five years.

The awards will fund 14 lead U.S. institutions and U.S. partner institutions for collaborative projects involving international partners in 24 countries: Argentina, Australia, Bolivia, Brazil, Canada, China, Colombia, Cuba, France, Germany, Italy, Japan, Malawi, Malaysia, Mexico, Netherlands, Norway, Peru, Philippines, Switzerland, Taiwan, United Kingdom, Zambia and Zimbabwe.

“By linking together researchers from around the world, PIRE allows us to leverage U.S. dollars and improve scientific outcomes,” said Rebecca Keiser, head of NSF’s Office of International Science and Engineering (OISE), which manages PIRE. “These rich partnerships tackle some of today’s most pressing research questions, from new materials to marine sciences.”

NSF has a long history of fostering and supporting international relationships to address critical science and engineering (S&E) questions. Since its inception in 2005, the PIRE program has accelerated scientific discovery and enhanced the U.S. science and technology workforce by leveraging investments from foreign governments that also provide funding to these collaborative projects.

PIRE supports fundamental, international research and education in physical, living, human and engineered systems. As the focal point for international collaboration across NSF, OISE, which funds the U.S. portion of the international collaboration, catalyzes global S&E activities and builds effective partnership throughout the international S&E research and education community.

The new PIRE awards are briefly described below:

Hybrid Materials for Quantum Science and Engineering (HYBRID): Sergey Frolov, University of Pittsburgh. Partner country: France.

This project will study the effect of crystal growth and fabrication of quantum devices based on newly synthesized materials aided by theoretical and computational studies.

Bio-inspired Materials and Systems: LaShanda Korley, Case Western Reserve University. Partner country: Switzerland.

The project will use biomimetic materials to design adaptive and/or tough materials, implantable materials, excitable fibers and gels, and dynamic and functional fibers for soft robotic applications.

Science of Design for Societal-Scale Cyber-Physical Systems: Janos Sztipanovits, Vanderbilt University. Partner country: Germany.

This project seeks to develop a new Science of Design for societal-scale Cyber- Physical Systems such as traffic networks, electric grids, or networks of autonomous systems (e.g. self-driving cars, unmanned air vehicles) where control is dynamically shifted between humans and machines.

Investigation of Multi-scale, Multi-phase Phenomena in Complex Fluids for the Energy Industries: Masahiro Kawaji, CUNY City College. Partner countries: France, Germany and Norway.

This project will investigate multiscale phenomena in complex, multiphase fluids vital to energy technologies, such as the production and processing of oil and gas, energy conversion and storage, refrigeration and heating and cooling.

High Temperature Ceramic Fibers: Polymer-Based Manufacturing, Nanostructure, and Performance: Gurpreet Singh, Kansas State University. Partner countries: France, Germany Italy, and Japan.

This project will advance the materials science of certain polymer-derived ceramic fibers to reduce costs and improve performance for high temperature applications, particularly jet aircraft turbines.

Centennial Genetic and Species Transformations in the Epicenter of Marine Biodiversity: Kent Carpenter, Old Dominion University Research Foundation. Partner country: Philippines.

This project will compare the genetic diversity of fish collected in the Philippines during the 1907-1909 expedition of the U.S. Bureau of Fisheries ship Albatross with the same location’s present population to examine the loss of genetic diversity.

Advancing global strategies and understanding on the origin of ciguatera fish poisoning in tropical oceans: Alison Robertson, University of South Alabama. Partner countries: Australia, Canada, China, Cuba, Malaysia, Norway, Philippines and United Kingdom.

The project will investigate the threat to coral reef ecosystems by ciguatera fish poisoning, the most common nonbacterial seafood illness. This project will extend understanding of the environmental conditions affecting the production of ciguatoxins, and determine the fate of the toxins through the food web across geographical regions.

International partnership for cirrus studies: Elizabeth Moyer, University of Chicago. Partner countries: France, Germany and Switzerland.

This project will study one of the coldest and least understood parts of the atmosphere, the tropical tropopause layer (TTL). New satellite measurements suggest the TTL ice crystal layer is denser than previously thought, resulting in a stronger effect on the Earth’s radiation level.

Climate Research Education in the Americas using tree-ring and cave sediment examples: Mathias Vuille, University at Albany-State University of New York. Partner countries: Argentina, Bolivia, Brazil and Peru.

This project will investigate the nature of extreme events over the Americas during the last one thousand years by merging data from the two largest tree ring and cave sediment (stalagmite) archives in South America with new, unpublished records. This research will enhance scientists’ understanding of the underlying causes of past climate perturbations.

Black Hole Astrophysics in the Era of Distributed Resources and Expertise: Dimitrios Psaltis, University of Arizona. Partner countries: Germany, Mexico and Taiwan.

This project will fund activities in detector development, mounting coordinated observations, fast data transfer and cloud computing for the Event Horizon Telescope (EHT), an Earth-sized array of telescopes. The EHT seeks to capture the first images of astrophysical black holes and test Einstein’s theory of general relativity in extreme conditions.

Advanced Germanium Detectors and Technologies for Underground Physics: Dongming Mei, University of South Dakota. Partner countries: Canada, China, Germany and Taiwan.

This project will develop germanium material platforms used for large scale dark matter and neutrinoless double-beta decay experiments. The nature of dark matter and the key properties of neutrinos are two of the most important questions in fundamental physics.

Computationally-Based Imaging of Structure in Materials (CuBISM): Kenneth Shull, Northwestern University. Partner countries: France, Italy and Netherlands.

This project will develop computational and experimental tools to understand property degradation over long periods of time by examining historic art objects.

Confronting Energy Poverty: Building an Interdisciplinary Evidence Base, Network, and Capacity for Transformative Change: Pamela Jagger, University of North Carolina at Chapel Hill. Partner countries: Malawi, Zambia and Zimbabwe.

This project will test the impact of energy poverty mitigation interventions on air quality, land use and human welfare. Results will help address the resulting negative impacts on environmental sustainability and human health.

PIRE-Sustainable Communities & Gold Supply Chains: Integrating Responsible Engineering & Local Knowledge to Design, Implement & Evaluate Sustainable Artisanal Mining in Latin America: Juan Lucena, Colorado School of Mines. Partner countries: Colombia and Peru.

This project will study sustainability of artisanal and small-scale gold mining (ASGM). ASGM causes large-scale deforestation, air and water contamination, and chronic human diseases from the mercury used to process the ore.

About NSF

The National Science Foundation (NSF) is an independent federal agency that supports fundamental research and education across all fields of science and engineering. In fiscal year (FY) 2017, its budget is $7.5 billion. NSF funds reach all 50 states through grants to nearly 2,000 colleges, universities and other institutions. Each year, NSF receives more than 48,000 competitive proposals for funding and makes about 12,000 new funding awards.

Source: NSF

The post NSF Announces 14 New PIRE Awards to Support Scientific Collaboration in 24 Countries appeared first on HPCwire.