October 14, 2010
Here is a collection of highlights from this week's news stream as reported by HPCwire.
Teams Gear Up for SC10 Student Cluster Competition
Ixia's High Speed Ethernet Test Solutions Validate Mellanox ConnectX-2 NIC
Astronomer Employs HPC to Peer into Cosmic Mysteries
Nimbis Services Announces Cloud Services for Mathematica
Extreme Networks, SARA and CERN Complete First Long-Haul 40 GbE Link Across Europe
Ontario Advanced Network First to Launch Operational 100G Network
NOAA Establishes Supercomputing Center in West Virginia
Platform Computing, Excelian Partner to Deliver High Performance Analytics for Financial Services
AMAX Partners with Bright Computing to Maximize HPC Throughput
Ciena, Mellanox, SURFnet and University of Amsterdam Complete 40 Gbps, Long-Haul Optics Demonstration
Intel Reports First $11 Billion Revenue Quarter
SGI Announces Support and Record Benchmarks for VoltDB Database
Appro HF1 Server Targets High Frequency Traders
NEC Selects PBS Works as the Workload Management Tool for TSUBAME 2.0
Pittsburgh Supercomputing Center Boots Up 'Blacklight'
Nimbis Services Launches HPC Workbook for Desktop Users
Deterministic Parallel Java Brings Safety and Modularity to Parallel Programming
Multicore Processors Key to Increasing Flexibility in High-Performance Networking, Report Finds
BLADE Switch Delivers One Terabit of Throughput
Switch maker BLADE Network Technologies (BLADE) today unveiled the RackSwitch G8264, a single-chip 40 Gigabit Ethernet (GbE) top-of-rack switch. The switch delivers more than one terabit of low-latency throughput to the datacenter. This is the first time that a single-chip switch is available for terabit-scale deployment of 10GbE.
The new switch touts 64-10GbE ports, up to four-40GbE ports and 1.28 terabits of non-blocking throughput. Designed to handle I/O-intensive and highly virtualized workloads, the switch is well suited for HPC clusters, cloud computing, and algorithmic trading.
BLADE is aiming to fulfill the needs of mainstream enterprise datacenters, which are responding to increased data demands by increasingly deploying servers equipped with 10GbE. BLADE is going forward with the belief that 40GbE is the next logical step. Higher speed uplinks, such as 10/40 Gigabit Ethernet switches, will be required to handle the increased network bandwidth of the next-generation of datacenters.
According to Vikram Mehta, president and CEO, BLADE Network Technologies:
"BLADE is proud to break the terabit barrier in a single-chip design with the RackSwitch G8264. Our new switch is designed for today's most demanding requirements at the datacenter edge to interconnect highly utilized servers equipped with 10 Gigabit Ethernet and provide seamless migration to 40 Gigabit upstream networks."
The RackSwitch G8264 will be available in November at a cost of $22,500 USD. Interested parties can view the product at the upcoming Supercomputing Conference (SC10).2
UC Riverside Physicists Advance Spin Computing
"Spin computing" -- aka "spintronics" offers great potential for the future of computing -- think superfast computers that can overcome present Moore's Law limitations while using less energy and generating less heat than the current batch of number crunchers.
Here's how it works: electrons can be polarized so that they have a particular directional orientation, called spin. An electron can either be polarized so attain two states, called "spin up" or "spin down." Storing data with spin would effectively double the amount of data a computer could store since it allows two pieces of data to be stored on an electron instead of just one, as is currently the case.
While researchers have been working on the technology for about four decades, it's not quite ready for primetime. This week, however, Physicists at the University of California, Riverside have taken spintronics to the next level by successfully achieving "tunneling spin injection" into graphene. Their study results appear this week in Physical Review Letters.
From the announcement:
Tunneling spin injection is a term used to describe conductivity through an insulator. Graphene, brought into the limelight by this year's Nobel Prize in physics, is a single-atom-thick sheet of carbon atoms arrayed in a honeycomb pattern. Extremely strong and flexible, it is a good conductor of electricity and capable of resisting heat.
While graphene has characteristics that make it a very promising candidate for use in spin computers, the electrical spin injection from a ferromagnetic electrode into graphene is inefficient. Additionally, and even more troubling to the research team, observed spin lifetimes are thousands of times shorter than expected theoretically. Longer spin lifetimes are important because they allow for more computational operations.
The research team, led by Roland Kawakami, an associate professor of physics and astronomy, was able to dramatically increase the spin injection efficiency by inserting an insulating layer, known as a "tunnel barrier," in between the electrode and the graphene layer. The team thus achieved the first demonstration of tunneling spin injection into graphene, and the 30-fold increase spin injection efficiency set a world record.
The Kawakami lab was also to reconcile the short spin lifetimes of electrons in graphene. They discovered that using the tunnel barrier increased the spin lifetime. According to Kawakami, graphene has the potential for extremely long spin lifetimes.
The next step for the Kawakami lab is to demonstrate a working spin logic device. Ultimately, a chip capable of manipulating the spin of a single electron could pave the way for futuristic quantum computers.
10/30/2013 | Cray, DDN, Mellanox, NetApp, ScaleMP, Supermicro, Xyratex | Creating data is easy… the challenge is getting it to the right place to make use of it. This paper discusses fresh solutions that can directly increase I/O efficiency, and the applications of these solutions to current, and new technology infrastructures.
10/01/2013 | IBM | A new trend is developing in the HPC space that is also affecting enterprise computing productivity with the arrival of “ultra-dense” hyper-scale servers.
Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?