Today Samsung Electronics announced that it has begun mass producing the industry’s first 4-gigabyte DRAM package based on the second-generation High Bandwidth Memory (HBM2) interface, for use in high performance computing, advanced graphics and network systems, as well as enterprise servers. Samsung’s new HBM solution will offer unprecedented DRAM performance – more than seven times faster than the current DRAM performance limit, allowing faster responsiveness for high-end computing tasks including parallel computing, graphics rendering and machine learning.
Today Allinea announced that Oak Ridge National Laboratory has deployed its code performance profiler Allinea MAP in strength on the Titan supercomputer. Allinea MAP enables developers of software for supercomputers of all sizes to produce faster code. Its deployment on Titan will help to use the system’s 299,008 CPU cores and 18,688 GPUs more efficiently. Software teams at Oak Ridge are also preparing for the arrival of the next generation supercomputer, the Summit pre-Exascale system – which will be capable of over 150 PetaFLOPS in 2018.
Although liquid cooling is considered by many to be the future for data centers, the fact remains that there are some who do not yet need to make a full transformation to liquid cooling. Others are restricted until the next budget cycle. Whatever the reason, new technologies like Internal Loop are more affordable than liquid cooling and can replaces less efficient air coolers. This enables HPC data centers to still utilize the highest performing CPUs and GPUs.
The 2016 OpenFabrics Workshop has extended the dealing for its Call for Sessions to Feb. 1, 2016. The event takes place April 4-8, 2016 in Monterey, California. “The Workshop is the premier event for collaboration between OpenFabrics Software (OFS) producers and those whose systems and applications depend on the technology. Every year, the workshop generates lively exchanges among Alliance members, developers and users who all share a vested interest in high performance networks.”
Although the cloud has become an accepted part of commercial and consumer computing, science and engineering have been less welcoming to the concept, but this could be on the point of changing with the announcement this month that the ESI Group will be delivering advanced engineering simulation in the cloud, across multiple physics and engineering disciplines.
In this video from the Intel HPC Developer Conference at SC15, Kevin O’Leary from Intel presents: Vectorization Advisor in Action for Computer-Aided Formulation. “The talk will focus on a step-by-step walkthrough of optimizations for an industry code by using the new Vectorization Advisor (as part of Intel® Advisor XE 2016). Using this tool, HPC experts at UK Daresbury Lab were able to spot new SIMD modernization and optimization opportunities in the DL_MESO application – an industry engine currently used by “computer-aided formulation” companies like Unilever.”
The 32nd International Conference on Massive Storage Systems and Technology (MSST 2016) has issued its Call for Participation & Papers. The event takes place April 30 – May 6 in Santa Clara, CA. “The Program Committee requests presentation proposals on issues in designing, building, maintaining, and migrating large-scale systems that implement databases and other kinds of large, typically persistent, web-scale stores (HSM, NoSQL, key-value stores, etc.), and archives at scales of tens of petabytes to exabytes and beyond.”
Today Intersect360 Research released its eighth 2015 Site Budget Allocation Map, a look at how HPC sites divide and spend their budgets.
“Computers are an invaluable tool for most scientific fields. It is used to process measurement data and make simulation models of e.g. the climate or the universe. Brian Vinter talks about what makes a computer a supercomputer, and why it is so hard to build and program supercomputers.”
Today Baidu’s Silicon Valley AI Lab (SVAIL) released Warp-CTC open source software for the machine learning community. Warp-CTC is an implementation of the #CTC algorithm for #CPUs and NVIDIA #GPUs. “According to SVAIL, Warp-CTC is 10-400x faster than current implementations. It makes end-to-end deep learning easier and faster so researchers can make progress more rapidly.”