|
| This Article | ||
| ||
| Share | ||
| Bibliographic References | ||
| Add to: | ||
 Digg  Furl  Spurl  Blink  Simpy  Google  Del.icio.us  Y!MyWeb | ||
| Search | ||
| ||
| ASCII Text | x | ||
| Samar Sharma, Rina Panigrahy, "Sorting and Searching Using Ternary CAMs," High-Performance Interconnects, Symposium on, pp. 101, 10th Symposium on High Performance Interconnects HOT Interconnects (HotI'02), 2002. | |||
| BibTex | x | ||
| @article{ 10.1109/CONECT.2002.1039264, author = {Samar Sharma and Rina Panigrahy}, title = {Sorting and Searching Using Ternary CAMs}, journal ={High-Performance Interconnects, Symposium on}, volume = {0}, year = {2002}, isbn = {0-7695-1650-5}, pages = {101}, doi = {http://doi.ieeecomputersociety.org/10.1109/CONECT.2002.1039264}, publisher = {IEEE Computer Society}, address = {Los Alamitos, CA, USA}, } | |||
| RefWorks Procite/RefMan/Endnote | x | ||
| TY - CONF JO - High-Performance Interconnects, Symposium on TI - Sorting and Searching Using Ternary CAMs SN - 0-7695-1650-5 SP EP A1 - Samar Sharma, A1 - Rina Panigrahy, PY - 2002 KW - null VL - 0 JA - High-Performance Interconnects, Symposium on ER - | |||
Sorting and searching are classic problems in computing and several RAM based solutions exist. In this paper, we present sorting and searching algorithms using TCAMs. Using these algorithms, one can perform sorting in O(n) memory cycles using a TCAM. Furthermore, inserts and deletes to a sorted list and priority queue operations can be done in O(1) cycles.
A searching problem may typically involve finding if a query point is contained in a given set of ranges. We call this the point intersection problem. The ranges may be inserted or deleted from the set dynamically.
For these problems, we present several algorithms with different time, space and implementation complexity tradeoffs. One of the algorithms uses only O(1) time for all operations and O(1) space for each range in the set when the ranges are disjoint. More precisely, one can perform disjoint range search in one memory cycle using a 2-stage pipeline. We also provide several solutions in scenarios where the ranges in the set are allowed to overlap.