Cisco is not currently accepting proposals for this RFP.
Project ID:
RFP-2007-019
Title:
Methods for Developing Efficient Multicore Algorithms
Summary:
Recent advances in multicore architectures require a redesign of many fundamental applications to better fit multiprocessors which have small instruction and data caches. Current algorithms for various applications such as deep packet inspection, regular expression matching, anomaly detection and specialized hash function computations are optimized either for hardware implementations or for expensive (in cost and energy consumption) general purpose CPUs. The prevalence of multicore machines with small L1 cache and shared L2 cache calls for new methodologies for efficient design. The question is, can we develop methods for constructing algorithms that are well-suited to multicore cache architectures, without compromising algorithm generality or performance?
Full Description:
Contemporary networking equipment must sustain extremely high throughput while using cheap multicore hardware. Features which until recently were either implemented in hardware (such as hash function computations), or on powerful general purpose machines (such as regular expression matching) must now be implemented on cache-starved multicore machines. The challenge is to find new paradigms and design patterns to fully exploit the core parallelism while minimizing contention for shared memory bus and L2 cache resources.
We are interested in algorithms that take into account cache specifics, without being limited to a particular processor architecture. For example, such algorithms could balance the cost of cache misses with the cost of CPU computation. Applications ranging from hash function computations, to regular expression matching and XML parsing could benefit from such algorithms.
Constraints and other information:
IPR will stay with the University. Cisco expects customary scholarly dissemination of results, and hopes that promising results would be made available to the community without limiting licenses, royalties, or other encumbrances.
Proposal submission:
Cisco is not currently accepting proposals for this RFP.
Questions? Contact: research@cisco.com
