Silicon Valley Engineering Council
Thanks to Our Sponsors
SVEC 2014 Open House Reception Friday, November 21, 2014
with talk on Cost-Effective Supercomputing with Highly Parallel Systems
SVEC is partnering with Paypals TechXploration meetup for this event. Please join the Silicon Valley Engineering Council (SVEC) at our annual Open House event Friday November 21st to share contacts, ideas and engineering camaraderie. Join us at PayPal TownHall for an exciting evening. Colleagues, friends and spouses are invited.
5:30pm The event begins with a networking reception and hors doeuvres
Highlights of the Open House program will include:
Networking Reception with many local engineers
How to find us: TechXploration takes place at PayPal Town Hall (2161 North 1st Street, San Jose, CA. Map to PayPal TownHall), which is facing North First Street. It is directly across the street from the Karina Station of VTA/Light Rail in San Jose, CA.
Cost-Effective Supercomputing with Highly Parallel Systems
Todays fast-paced, highly integrated, product development environment, demands that products be both created and tested in computer simulated environments. Finite Element Analysis (FEA), the choice for most simulations, is used for calculations of stress, heat transfer, electrostatics, electromagnetic fields, fluid mechanics, and many more topics. Accurate calculations results with fine resolution place a heavy demand on the computational resources, especially for coupled problems. Problems with fine resolution or complex temporal development can require formidable supercomputing resources, but most commercial FEA packages are not configured for massively parallel operations, are generally expensive, and none solve all problem types within the same package. Commercial FEA packages also tend to be limited in that they generally work within a shared memory server, (usually one workstation), and are not scalable to massive parallelism. Today we present a highly parallel FEA approach, which uses a combination of open source software, public domain mathematics libraries built by National Laboratories, and a variety of matrix solvers. Our approach utilizes a distributed system that fully utilizes a rack of servers that offer all their RAM and processors/threads to the mother process (node). The parallel approach nears zero latency on even a mere Gigabit Ethernet network, thereby allowing, during a solution cycle, all hyper-threads (over all servers) to be running at close to 100% capacity, all the time. Relatively inexpensive rack servers can be stacked and networked to create a system that acts like it has thousands of GB of RAM and hundreds, if not thousands of processor cores. This scalability is linear, in that processing speed is proportional to the number of servers (RAM and processors) in the system. This approach brings immediate near-term benefits, and promises making cost-effective massively-parallel computing generally available in the commercial sector.
Bio of David Eimerl: http://www.eimexinc.com/meetdavid.html
Location: PayPal TownHall
The Silicon Valley Engineering Council, the alliance for engineering leaders in the Silicon Valley, serves as a collective focus to support your professional engineering organization and to serve the needs of your membership and the engineering education community in the San Francisco Bay Area.
Maker Camp for kids/teens
Elise Engelhardt is putting on a Maker Camp for kids/teens every Wednesday 2pm this summer (for six weeks). Maker Faire has donated materials including several Arduinos. If you can help with some of the workshops please sign up as a mentor here: