IEEE Micro, March/April 2009, pp. 2–3
Software Standards for the Multicore Era
Pushing single-thread performance is no longer viable for further scaling of microprocessor performance, due to this approach's unacceptable power characteristics. Current and future processor chips will comprise tens to thousands of cores and specialized hardware acceleration to achieve performance within power budgets.4-7 Hardware designers find it relatively easy to design multicore systems. But multicore presents new challenges to programmers, including finding and implementing parallelism, dealing with debugging deadlocks and race conditions (that is, conditions that occur when one or more software tasks attempt to access shared program variables without proper synchronization), and eliminating performance bottlenecks.
Theme: WEB 2.0
What else is new?
Biologically Inspired Computing
Evolutionary algorithms and robotics hold great promise as integrated design and modeling tools.
Building Broadband as Economic Stimulus
Learn about the broad ideals motivating the small part of the US$787 billion stimulus bill devoted to building more broadband and wireless services.
Keys Don’t Grow in Threes
Many Internet security mechanisms depend on cryptographic algorithms for authentication and confidentiality, but some parameters must be specified, prompting calls for widespread changes by 2010.
Robotic Companions for Smart Space Interactions
Emerging technology uses robotic companions as interfaces in smart spaces. A case study examines social robots in the Ubiquitous Home at NICT, Japan.
The Story behind the Intel Atom Processor Success
This article analyzes the Intel Atom processor from a management perspective as an example of effective project execution.
The Real Cost of Software Errors
Modern society's reliance on software even in critical infrastructure systems makes software failures difficult to ignore. The cost of such
failures manifests itself not only in dollars but also in lives.
Strategies and Policies to Support and Advance Education in e-Science
With the appropriate support, e-Science can help improve research, innovation, decision support, or diagnosis in any discipline.