insideBIGDATA : In the Fall of 2013, Harvard University offered CS109 Data Science which is an excellent introductory course for those interested in getting a jump start into this exciting field. Most of the class materials including video lecture archives and slides are freely available online. This is a fantastic way to get ivy-league quality education, albeit without university credit. The course is currently taught by two Harvard professors: Hanspeter Pfister (Computer Science) and Joe Blitzstein (Statistics).
This course introduces the following aspects of data science:
- Data munging, cleaning, and sampling
- Data management to be able to access big data quickly and reliably
- Exploratory data analysis to generate hypotheses and intuition
- Prediction based on statistical methods such as regression and classification
- Communication of results through visualization, stories, and summaries
The GPU Technology Theater features talks on a wide range of topics in HPC and accelerated computing :
From November 17 to 22 Shanghai Jiao Tong University was present at the biggest supercomputing event of the world through the Center of High Performance Computing. The representatives were James Lin, head of the center, and Éric Germaneau.
We explained the specificity of the new supercomputer of SJTU, л. The posters on the back detailed the center as well as the technical details of the cluster. Many people came to visit our booth, in such we spread the information that SJTU is present in the HPC word. Contacts and exchanges were very productive. As a consequence several opportunitiesof collaboration may be created in a near future.
The conference welcomed more than ten thousand participants and attendees, it was a wonderful source of knowledge. It is very important to participate to such event for what we get from it let us provide a better service to our usersand also make a teaching that follow the technology. Technology develops extremely fast now days and participating to such international conference allow us to have a teaching content that is not out of date. This is a unique opportunity to know about the latest technology,our students deserve to learn recent and useful things.
HPCwire: While we hope you’ll stay tuned this morning for a detailed analysis of major trends on this year’s Top500 list, for now, we wanted to deliver the breaking results of this year’s systems, many of which are not surprising when compared to last year’s June list. As you can see below from the Top 10, very little movement–although there are some notable trends on the processor, interconnect, accelerator and other fronts that we will discuss later this morning.
The only surprise in the top ten this year is Piz Daint, a Cray XC30 system installed at the Swiss National Supercomputing Centre (CSCS) in Lugano, Switzerland, which now holds the title of the most powerful system in Europe. Piz Daint achieved 6.27 Pflop/s on the Linpack benchmark. Piz Daint is also the most energy efficient system in the TOP10 consuming a total of 2.33 MW and delivering 2.7 Gflops/W
insideHPC : Today Nvidia announced CUDA 6, the latest version of the company’s pervasive parallel computing platform and programming model. Designed to make parallel programming easier than ever, Cuda 6 will help developers decrease the time and effort required to accelerate their scientific and engineering applications with GPUs.
By automatically handling data management, Unified Memory enables us to quickly prototype kernels running on the GPU and reduces code complexity, cutting development time by up to 50 percent,” said Rob Hoekstra, manager of Scalable Algorithms Department at Sandia National Laboratories. “Having this capability will be very useful as we determine future programming model choices and port more sophisticated, larger codes to GPUs.”
Key features of CUDA 6 include: 阅读全文…
insideBIGDATA : In a significant coupling of scientific research and big data, NASA and Amazon Web Services Inc. (AWS) are making a large collection of NASA climate and Earth science satellite data available to research and educational users through the AWS cloud. The system enhances research and educational opportunities for the U.S. geoscience community by promoting community-driven research, innovation and collaboration.
insideBIGDATA : SriSatish Ambati from 0xdata describes the company’s new H20 Open Source, In-memory Machine Learning application for Big
In the midst of rapid development of high performance computing beyond the petascale and the emergence of new scientific and engineering applications both in China and worldwide, this conference will highlight the current research progress in scientific computing in terms of
Shanghai Jiao Tong University (SJTU) officially put its new supercomputer into production on Wednesday 23 of October. The freshly launched machine, called π, is the most powerful HPC cluster among all the universities in China and was ranked 158 of the TOP500 in June 2013. The π is hosted and managed by the Center for High Performance Computing. It is composed by the latest technology available today and includes 5312 Xeon SNB E5-2670 processor cores, 100 Nvidia Kepler K20 cards and 10 Intel Xeon Phi 5110P coprocessors. This allows the cluster to reach the theoretical peak performance of 196.2 Tflops. Each of the 412 Intel EPSD JP/WP compute nodes are interconnected by the Infini Band Mellanox network and its liquid cooling System is provided by Rittal Liquid. The machine also supports 720TB of disc space as well as 80 Intel SSD providing 400G of memory.
The π is also a unique example of partnership since no less than three different companies joined to design and build it. The cluster was jointly developed by Nvidia, Intel, Inspur.
The π is aimed to provide high performance computing service to students and academics. A wide range of applications are running on π that include computer science, computational material science, computational physics and chemistry, biosimulation, and big data analysis. Researchers can perform large-scale or multi-scale simulations and more deeply explore their fields.
Professor Sun, of the Computational Chemistry group, School of Chemistry and Chemical engineering (SJTU), explained what he expects from the computational power of the cluster: π will help them to know the “real limits of simulation of time and space scale” and “discover new phenomena”.
Udacity online GPU course is now on youku: