So here is how I'm going to chronicle this eruption of news and information. First, I should mention that Joe Landman is blogging as well, he also has pictures. Check his site as well this site for news -- clearly too much for any one site.
For the record, I'm helping out at the Linux Magazine booth (1815) at the the show. If you are at the show and want a free copy (and two more by mail) please stop buy. There is a great article about a $2500 cluster and optimizing the HPL benchmark written by cough, cough, ahh ... there are other good articles as well.
So here is how I'm going to do this. I'll put together a page for each day and also have news releases and announcements. I'll run the announcements like a push down stack, newest go on top.
- SC05 Monday
- SC05 Tuesday - Jeff's Monkey Blog Day 1
- SC05 Wednesday - Jeff's Monkey Blog Day 2
- SC05 Thursday - Jeff's Monkey Blog Day 3
Well, I (Doug), never really go to blog about anything. I found that there were plenty of great conversations, talks, BOFs, and events that sucked up my time more than I had imagined. The Monkey Mingle on Monday night was well attended and we manged to give out all of the Cluster Monkey hats. All those who received hats have agreed to wear them for one year, by the way. Tuesday was the Bill Gates talk, which I will address in a separate article. After having dinner with some of the fine people from the Aggregate.org and the Warewulf Project, I did manage to see the last three songs by Sheryl Crow at the Microsoft party (yes that is right Sheryl Crow). On Wednesday, I was torn between seeing Steven Hawking (appearing down the street) or attending the Beowulf Bash. It was a tough call -- Quantum physics and enlightenment or hanging out with a bunch of drunk cluster hackers. Well, those who know me, understand my quest for higher meaning, which I guess is why I decided to go with the drunk cluster hackers option. We finished the evening (i.e. got kicked out of) at some sports bar near the convention center where we were holding a LECCIBG. How the heck are we going to move this market forward with attitudes like that. And, just in case anyone was wondering, I manged to see Steven Hawking in another universe, unfortunately, I collapsed in the one with the cigars.
Overall, there did not seem to be any BIG breakthrough news from SC05. Clusters are winning big as is Linux. The Top500 had a bit of a shake-up, but no big surprises. If you get past the contest part, there are lots of interesting historical trends that are worth a look. Consider the growth of Linux and clusters over the last 5 years. Not that this is big news to the regular Cluster Monkey readers. I believe this is what is called a disruptive technology in some circles. I also found the steady increase of industrial HPC a welcome trend. While you are at the Top500 site, check out the other trends as well.
At some point, we have to mention the Microsoft foray into HPC. Indeed, it actual warrants a separate article.
AnnouncementsMyricom and Extreme Networks Target High Performance Computing and Data Centers with Proven 10-Gigabit Solutions Myricom, Inc. and Extreme Networks, Inc., the leader in open converged networks, are teaming up to help customers with High-Performance Computing (HPC) and mainstream networking applications with jointly tested network solutions that achieve 10-Gigabit performance. The two companies demonstrated interoperability in booth #436 at the SC|05 event held in Seattle. Upon entering an agreement where Myricom has joined Extreme Networks'(R) Go Purple Partner Program, the companies conducted a comprehensive series of tests combining high-availability switches from Extreme Networks with Myricom's Myri-10G Network Interface Cards (NICs) and achieved wire-speed performance at 10-Gigabit speeds.
New Moab Grid SuiteÂ® advances resource control, sovereignty & usability for real-world grid environments Cluster Resources, Inc. announced the release of Moab Grid Suite 4.5, its new grid management solution that lets organizations consolidate management of multiple clusters to create usable, affordable grids while maintaining local cluster sovereignty. OpenMPI 1.0 relased Open MPI is a project combining technologies and resources from several other projects (FT-MPI, LA-MPI, LAM/MPI, and PACX-MPI) in order to build the best MPI library available. A completely new MPI-2 compliant implementation, Open MPI offers advantages for system and software vendors, application developers and computer science researchers.
AMD and Sun Microsystems Collaborate to Build Japan's Largest Supercomputer for Tokyo Institute of Technology - AMD and Sun Microsystems, Inc. announced that the Tokyo Institute of Technology, one of the world's leading technical institutes, is creating Japan's largest supercomputer on a foundation of Sun. The system is based on Sun Fireâ¢ x64 (x86, 64-bit) servers with 10,480 AMD Opteronâ¢ processor cores [totaling more than 50 trillion floating point operations per second (teraFLOPS)], Sun and NEC storage technologies and NECâs integration expertise as well as ClearSpeed SIMD acceleration boards.
Linux Networx to Focus on Delivering Linux Supersystems Linux Networx, The Linux Supercomputing Company, announced the availability of the LS Series, a family of Linux Supersystems that blend the price and performance of Linux clusters with the power of legacy supercomputing, in a high value supercomputing system. The LS Series makes Linux supercomputing available to a broader range of users thanks to a combination of optimized performance and reduced Total Cost of Ownership. LS Series Supersystems are delivered ready for Performance on Power-up, eliminating the cost and complexity associated with first generation supercomputing clusters. You can read more here.
Penguin Computing Unveils Application-Ready Clusterâ¢ Penguin Computing today unveiled the Penguin Application-Ready Clusterâ¢ Portfolio, a new line of integrated hardware/software Linux-based cluster systems that provides greater productivity and efficiency in high performance computing (HPC) and commercial data center environments. The pre-configured, pre-tested Application-Ready Cluster Portfolio reduces time to deployment and eliminates the need for many traditional cluster end user and system administration tasks, making clustered computing dramatically simpler and more efficient. This is cool.
PathScale InfiniPath Powers One of the World's Largest Dual-Core AMD Opteronâ¢ Processor-Based InfiniBand Clusters PathScale, the developer of innovative software and hardware solutions to accelerate high performance computing, today announced that its InfiniPathâ¢ HTXâ¢ InfiniBandâ¢ Adapters have been deployed by AMD in its Developer Center to maximize application scaling on the newly installed Dual-Core AMD Opteronâ¢ processor-based 'Emerald' cluster. The combination of Dual-Core AMD Opteron processors and the InfiniPath interconnect is already demonstrating unprecedented performance, enabling Emerald to outperform traditional supercomputers in several critical benchmarks in the latest High Performance Computing (HPC) Challenge.
MicrowayÂ® Delivers FasTreeâ¢ - The Fastest InfiniBand Switches Available Today, Incorporating Mellanox DDR Technology Microway is now shipping its FasTreeâ¢ family of InfiniBand switches. The initial FasTree offering is a 36-port DDR switch based on Mellanox's 24-port InfiniScale III 4X component. Up to this point, all InfiniBand links ran at 2.5 GHz. The new DDR switch announced this week uses copper links running at twice this speed, and is the first offering to the HPC community of an InfiniBand switch that uses DDR links. "The main benefits of FasTree technology are its distributed modularity, extremely low latency and extensibility. These features double the size of the three hop fat tree fabrics that can be created with it, while at the same time doubling the size of local low latency domains," commented Stephen Fried, Microway's co-founder and CTO.