The Leading Source for Global News and Information Covering the Ecosystem of High Productivity Computing
November 09, 2007
Fulcrum Targets 'Datacenter Ethernet' with New Switch
Fulcrum Microsystems has introduced its latest 10 Gigabit Ethernet switch silicon, the FM4000 series IP router chips. The company is positioning the new devices as the foundation of "Datacenter Ethernet," that is, Ethernet for high-performance computing, server and storage host interconnect, and datacenter aggregation. The FM4000 represents a superset of the FM2000 series, the company's first low-latency 10 GbE switch chip.
Like its FM2000 predecessor, the FM4000 supports up to 24 ports and a sub-microsecond chip latency, enabling fat-tree scalability to thousands of nodes. On top of the FM2000 capabilities, the FM4000 adds layer 3 IP routing, layer 4 TCP and UDP support, and converged enhanced Ethernet (CEE) capabilities. The integrated design means chip latency for the FM4000 is an impressive 300ns, just 100ns more than the 200ns for the non-routing FM2000. While current InfiniBand chip latencies are slightly less, at 140ns, Fulcrum has managed to get its silicon into the same ballpark.
So far, 10 GbE switching has been a no-show in the datacenter, but the folks at Fulcrum sees a big opportunity there. The holy grail of converged Ethernet in the datacenter would eliminate the current islands of computing that currently exist -- Fibre Channel for storage, Ethernet for the communication network between the datacenter tiers, and InfiniBand for HPC clusters.
The CEE capabilities promise to make Ethernet a more general-purpose solution for the datacenter. They provide the ability to pause communication traffic on a priority basis, which enables a lossless fabric implementation. Although CEE is not yet an Ethernet standard, Fulcrum wanted to provide some of these capabilities to give switch vendors the ability to push up against Fibre Channel and InfiniBand.
"There have been a lot of proprietary fabrics in the world, and those are all essentially dying out," said Mike Zeile, Fulcrum Microsystems vice president of marketing. "The last one standing is InfiniBand. It's our view that InfiniBand has worked out -- in the TOP500 at least -- in the absence of 10 Gig Ethernet. Once 10 Gig Ethernet becomes cost effective, along with low latency and improved scalability, then Ethernet wins."
Current implementations of 10 GbE switches using the FM2000 technology include the Quadrics QSTenG, the Force10 S2410 and the Woven EFX 1000. As yet, none of these has proved to be the "InfiniBand killer," although all have approached at least SDR IB switches in bandwidth and latency performance. The problem for Ethernet switch vendors is that InfiniBand is a moving target. The 10 GbE switches have just caught up with SDR, while InfiniBand vendors are already well into DDR (20 Gbps) deployment, with QDR (40 Gbps) coming next year. The additional features of the FM4000 chips, especially IP routing and CEE, may help to create some differentiation from InfiniBand solutions, but that remains to be seen.
We may not have to wait too long though. Coinciding with the Fulcrum announcement, startup Arastra is introducing its new 10 GbE switch based using the FM4000 silicon. The company has been in stealth mode while working in parallel with the Fulcrum team. Arastra is backed by a number of Silicon Valley veterans, including Sun Microsystems co-founder Andy Bechtolsheim.
Arastra's 7100 series 10 GbE switch is a high-density, 48-port top-of-rack switch that fits into a 1U box and can scale to thousands of ports. It uses the FM4000's CEE intelligence to manage congestion, prioritizing Ethernet traffic on a per port basis. The model enables much finer control of congestion management by slowing down the traffic at the source of the blockage.
"We envision that from the end of next year through 2009, most datacenter customers will be upgrading from 1 Gig to 10 Gig," said Mansour Karam, director of marketing for Arastra. "Our solution will enable this upgrade by offering dense, cost effective 10 Gig switching."
The 48-port switch can connect up to 40 servers and 8 uplinks. Early trials are ongoing with a number of government labs and Wall Street customers.
Article Tools
Share Options
(Digg, Technorati, more)
Subscribe
Sponsored Links
New Paper: Parallel Computing Without Parallel Programming
Learn how domain experts can run VHLL programs like MATLAB® on a variety of high-performance platforms without low-level reprogramming and how to work with the largest datasets and complex algorithms without sacrificing ease of use or reducing productivity.
RSS Feeds
Feeds by Topic
- Applications
- Developer Tools
- Interconnects
- Middleware
- Networks
- Processors
- Storage
- Systems
- Visualization
Feeds by Industry
Feeds by Content Type
Top Headlines
3D Seismic Data: Taking a Smarter Approach to Interpretation
Jul 09 | Engineer Live | The demand for computational tools to underpin the 3D seismic interpretation process has never been more apparent. Read more...
Engineering Unemployment Soared in 2Q to 8.6%
Jul 08 | EE Times | Unemployment for U.S. engineers has reached record levels, according to government figures. Read more...
Gartner Adjusts 2009 IT Spend Downward Again
Jul 08 | Network World | Global spending for 2009 projected to drop 6 percent, for a total of $3.2 trillion. Read more...
Concurrent and Parallel Are Not The Same
Jul 08 | Linux Magazine | Portability or efficiency? Neither is guaranteed when writing explicit parallel code. Read more...
800 TFLOP Real-Time Ray Tracing GPU Unveiled, Not for Gamers
Jul 07 | Ars Technica | Japanese company builds custom ASIC to accelerate real-time ray traced rendering for the auto industry. Read more...
Featured Whitepapers
Parallel Computing Without Parallel Programming
Jul 10 | | Engineers, scientists, and other domain experts depend on the productivity enabled by very high-level language (VHLL) tools like MATLAB® and Python. However, as datasets grow larger and programs get more sophisticated, ordinary desktop computers can no longer keep up. The paper explores how to run VHLL programs on high-performance platforms without low-level reprogramming. Work with large datasets and complex algorithms without sacrificing ease of use or reducing productivity.
Building High Performance Computing in a Green and Modular Solution Building Block
Apr 14 | | Many HPC IT departments are feeling the rising pressure to deliver more capacity computing and performance while trying to reduce the total cost of ownership. This white paper discusses how an environmentally-friendly and open-standards HPC building block based computing system using flexible interconnect options helps address capacity computing needs.
Multimedia
Webcast: Dell Expands HPC Access and Adoption with Intel Cluster Ready Program
Source: Addison Snell, GM/VP, Tabor Research; sponsored by Dell
Many organizations that could benefit from the use of HPC clusters find that it is complicated to get the systems up and running because of limited IT resources or the complexities of the clusters themselves. Learn how the Intel Cluster Ready program, for which Dell was an original partner, seeks to address this challenge for entry level and mid-range HPC users.
Video White Paper: Architecting a Better Network Storage Solution
BlueArc's Titan architecture represents an evolutionary step in file servers by creating a hardware-based file system that can scale bandwidth, IOPS, and overall data capacity well beyond conventional software-based devices. With its ability to virtualize a massive storage pool of up to four usable petabytes of tiered storage, Titan can scale with growing data requirements, offering a competitive advantage for businesses, researchers, or other enterprises seeking to better manage data growth while still ensuring optimal performance.
Webcast: HPC Development Solutions: Sun Studio & Sun HPC ClusterTools
Sun Studio Compilers and Tools and Sun HPC ClusterTools allow you to create high performance parallel applications for OpenSolaris, Solaris and Linux. Sun Studio Express 11/08 includes MPI performance analysis capabilities and full OpenMP 3.0 compiler support. Learn about all this and the latest in Sun HPC ClusterTools 8.1.
Newsletters
Stay informed! Subscribe to HPCwire email Newsletters.
HPC Job Bank
-
HPC Systems Administrator
McGill University -
High Performance Computing Systems Administrator
Idaho National Laboratory -
Solution Engineer- HPC
Dell Inc -
Parallel Programmer
Cardiff University -
Principal Software Research Engineer
Indiana University -
Post Doctoral Researcher - Exascale Stream Computing
IBM - Visit the HPCwire Job Bank
Featured Events
-
July 13-16, 2009
WORLDCOMP'09
Las Vegas , NV
USA
-
August 24-28, 2009
Data Mining: Levels I, II, & III
Denver , CO
USA