The dual data cache is a cache organization with a split temporal/spatial cache. The temporal sub-cache stores data exhibiting temporal locality and the spatial sub-cache saves data exhibiting spatial locality. A locality prediction table is used to predict the type of locality load/store instructions exhibit. In this way, both types of locality can be exploited more effectively. Unfortunately, the dual data cache does not make effective use of the entire cache capacity. If most memory references exhibit the same type of locality, only one sub-cache will be used. In this paper we, therefore, propose a cache organization called the Unified Dual Data Cache that employs only one (unified) cache unit. If a cache miss occurs and the locality prediction is temporal, only the missing block is fetched from the next memory level. If on the other hand spatial locality is predicted, adjacent blocks are also brought to the cache. In fact, we present two versions of the UDDC called the UDDC Type A (UDDC-A) and the UDDC Type B (UDDC-B), respectively. The difference between the two types is that in the UDDC-B each smaller block is tagged, while in the UDDC-A the smaller blocks within a larger block share the tag.