“With this award, Engility expands on our long-standing partnership with the FDA to further develop their high performance computing and complex data analysis capabilities,” said Engility CEO Lynn Dugle. “We look forward to supporting the FDA’s mission to make groundbreaking treatments, such as genetics-based medicine, available to all Americans.”
Are supercomputers practical for Deep Learning applications? Over at the Allinea Blog, Mark O’Connor writes that a recent experiment with machine learning optimization on the Archer supercomputer shows that relatively simple models run at sufficiently large scale can readily outperform more complex but less scalable models. “In the open science world, anyone running a HPC cluster can expect to see a surge in the number of people wanting to run deep learning workloads over the coming months.”
Growing momentum was the watchword at the inaugural OpenPOWER European Summit this week, where the OpenPOWER Foundation made a series of announcements today detailing the rapid growth, adoption and support of OpenPOWER across the continent. ” With today’s announcements by our European members, the OpenPOWER Foundation expands its reach, bringing open source, high performing, flexible and scalable solutions to organizations worldwide.”
Today Microsoft released an updated version of Microsoft Cognitive Toolkit, a system for deep learning that is used to speed advances in areas such as speech and image recognition and search relevance on CPUs and Nvidia GPUs. “We’ve taken it from a research tool to something that works in a production setting,” said Frank Seide, a principal researcher at Microsoft Artificial Intelligence and Research and a key architect of Microsoft Cognitive Toolkit.
Today Fujitsu Laboratories announced a collaboration with the University of Toronto to develop a new computing architecture to tackle a range of real-world issues by solving combinatorial optimization problems that involve finding the best combination of elements out of an enormous set of element combinations. “This architecture employs conventional semiconductor technology with flexible circuit configurations to allow it to handle a broader range of problems than current quantum computing can manage. In addition, multiple computation circuits can be run in parallel to perform the optimization computations, enabling scalability in terms of problem size and processing speed.”
“Over the past six weeks, we took NVIDIA’s developer conference on a world tour. The GPU Technology Conference (GTC) was started in 2009 to foster a new approach to high performance computing using massively parallel processing GPUs. GTC has become the epicenter of GPU deep learning — the new computing model that sparked the big bang of modern AI. It’s no secret that AI is spreading like wildfire. The number of GPU deep learning developers has leapt 25 times in just two years.”
In this video, Dr. Dimitri Kusnezov from the U.S. Department of Energy National Nuclear Security Administration presents: Supercomputing the Cancer Moonshot and Beyond. “How can the next generation of supercomputers unlock biomedical mysteries that will shape the future practice of medicine? Scientists behind the National Strategic Computing Initiative, a federal strategy for investing in high-performance computing, are exploring this question.”
Today Emu Technology announced that it has delivered an Emu Chick Memory Server to Oak Ridge National Laboratory. “ORNL intends to study the system for streaming graph analysis applications, sparse multilinear computations, and other memory-intensive problems, as we continue to test the potential of emerging computing technologies to further our mission within the DOE,” said Jeffrey S. Vetter, Director of the Future Technologies Group at ORNL’s Computer Science and Mathematics Division.
“This talk will provide empirical evidence from our Deep Speech work that application level performance (e.g. recognition accuracy) scales with data and compute, transforming some hard AI problems into problems of computational scale. It will describe the performance characteristics of Baidu’s deep learning workloads in detail, focusing on the recurrent neural networks used in Deep Speech as a case study. It will cover challenges to further improving performance, describe techniques that have allowed us to sustain 250 TFLOP/s when training a single model on a cluster of 128 GPUs, and discuss straightforward improvements that are likely to deliver even better performance.”
“The work we do involves capturing and analyzing huge environmental data sets so that the government can make informed policy decisions that protect humans and the environment,” said Ron Hines, Associate Director for Health at the EPA’s National Health and Environmental Effects Research Laboratory in Research Triangle Park, N.C. “We have collaborated with the NCDS on some of its initiatives in the past and having a seat at its leadership table will help us connect with leading data researchers, access data resources and infrastructure, and contribute to the development of future NCDS strategies.”