Much of the really exciting innovation in high-performance computing (HPC) is coming from accelerators, either in the form of GPGPUs or heterogeneous processors. Although there is growing vendor recognition of some of the issues that are specific to supercomputing, an ongoing consideration for the HPC community is that the economies-of-scale that drive commodity technology development do not always share the concerns of power, scalability and fault tolerance that are vital to the success of large-scale scientific computing efforts. This symposium will seek to identify these issues and discuss strategies for dealing with them. The resulting dialogue will be summarized to provide vendors with new information on how their technologies are being used and on what changes or additions might aid in greater adoption.
To register, visit www.gputechconf.com/page/registration-pricing.html.