Where will the new oil be discovered? What will be the impact on water resources with practices such as fracting, the process of extracting underground natural gas?

These are the types of topics that are being explored in the world of high performance computing.

I am live blogging a briefing at Intel Developer Forum on the topic of high performance computing.

11:12 a.m.: Berkeley professor discussing research in big data. Some stats: Facebook generates 200 to 400 terabytes of data a day with 83 million photos. In Berkeley’s amplab they use machine learning through algorithms, machines and people. The algorithms are for the HPC and cloud computing environments. The machines help with the crowdsourcing of the data and the people extract it.

For example, personalized medicine will allow for specific treatments and understandngs based on big data analysis. The need is there. About 25% of U.S. deaths are due to cancer. But what is cancer? It is the perversion of the normal cell. It is a genetic disease that can be triggered when exposed to carcinogens.

11:19 a.m.: In conjunction with the University of San Francisco, the researchers are using Hadoop and big data applications. Big question: will computer science trump medical researchers in helping discover cures?

11:24 a.m.: High performance computing is accelerating but there are complications. In particular is the issue of power. The Intel eon E5 processor is Intel’s answer. There is a 20x time demand compared to Xeon 5500.

11:29 a.m.: HPC is also relevant for commercial use but how to get to it? There is a lack of access. It is like a castle in a moat It is tough to access. The serfs, so to speak, fall into the moat trying to access. Digital development technology is part of the answer. The goal is to brak down the barriers.

11:31 a.m.: A representative from Lawrence Livermore Laboratory is here who is discussing the mission and how ti has broadened to a national security role. This is a multidisciplinary effort. Computing is now part of that mission.

The goal is to get answers to find better ways for using data in cyber security, energy security and nuclear counterterrorism to name a few.

11:45 a.m. Sorry, the site went down. The story here is about the problems in getting access to this kind of computing. You should not need a PHd to apply HPC. One solution is now write in Python and automatically parallelizing it to do computations. There are real issues with developing new languages, which will be needed for HPC. There is little chance it will succeed. So, the better use is to apply existing languages. Figure out how to expand it.

Services Angle