The Scalar Processors have a performance setback when it comes to Vector(Array) processing and other vector applications because of the Scalar processors limitation of processing single instruction at a time. The two most important reasons holding them back are: the use of same organization caching both spatial and temporal locality and the genrally implemented "eager" caching policy. This leads to the well known trade-off of designing caches with more number lines or implementing design which has large line size.
The remedy to such trade-off is a data cache organisation system called Dual Data Cache. Implementaion of such cache design leads to efficient exploitaion of both spatial and temporal localities in an efficient manner. The Dual Data Caching technique implement lazy caching technique which is a caching strategy that loads data into the cache when it is mandatory and that is achieved by the use of locality prediction table which maintains a history of details regarding the most recently executed load/store instruction.