Primeur magazine

Edition: flash - Issue: 2016-07-13

Industry news

In "What the FLOP/s? Part 2 - It Takes an Ecosystem!" Alex Larzelere summarises why in his view maintaining leadership in high performance computing is important for the US. "It took over 50 years for the U.S. to develop the ecosystem that once dominated the world of computing. National and economic security provides the reason the United States made the necessary investment to build an HPC ecosystem. Now in about 10 years it has been matched by China", he notes, and in conclusion: "As history has shown us, leadership in HPC translates into leadership producing the aircraft, automobiles, oil and gas, biomedical, military technology that continue to help the U.S. compete and win." Read further...
On the HPE blog, Bill Manel talks about high-frequency trading (HFT) technology. He notes: "In an effort to accelerate HFT, high performance computing (HPC) together with Big Data insights are being used to expedite transactions and promote data optimization. Successful financial organizations are investing in end-to-end computing solutions that can execute hundreds of trades per millisecond, including bundles of hardware and software that offer high-speed processing capabilities - an IT transformation that allows organizations to streamline operations, increase trader productivity, and boost network performance." Read further...
Thomas Skordas, Acting Director since July 1st of the renamed "Digital Excellence & Science infrastructure" Directorate of the European Commission summarizes in his blog the the achievements of the EC in the area of the European Open Science Cloud, the Quantum Computing Initiative, and the new "High Performance Computing & Quantum Technology" unit. Read further...
v2.0.0 is a major new release series containing many new features and bug fixes. Open MPI is now compliant with the MPI-3.1. standard. Open MPI RMA (Remote Memory Access) now maps to native RMA operations. Read further...

Exascale

Tobias Weinzierl from the ExaHype project wrote a report on the ISC 2016 workshop "Form Follows Function - Do algorithms and applications challenge or drag behind the hardware evolution?" Does co-design for exascale supercomputing really work? Most workshop participants were sceptical whether the cycle of influence is a good one the way we experience it right now: It orbits around weaknesses and demands. It is backward looking. Mark Parsons, EPCC, articulated that he is worried that the evolution even does not take the well-known Amdahl numbers into account: "I believe strongly in co-design but it happens extremely rarely." Read further...