Abstract: A cache memory apparatus allocates memory regions on the basis of information attributes. The required memory region corresponding to the attribute is accessed before caching is implemented. This enables considerable reduction in apparatus that would otherwise be required to be duplicated among different chips corresponding to the number of information attributes. In a multiple data processing implementation, cache memory regions are allocated for each data processor. The required memory regions of the cache memory are accessed in accordance with information specifying the data processor before caching is carried out. To avoid data collision when the main memory data is substantially rewritten, and when data having the same address and different access types exist simultaneously, data representing the type of access is stored. In response to a change in access, all data signals stored in each ways is nullified so that a plurality of data stored in the memory regions is not simultaneously output.

Abstract: In a data processor, when there is any cache memory not being activated after the whole data processor has been activated, a signal is delivered to a bus driver and then a data processing unit is connected to a system bus. During the period from when the whole data processor has been activated to when all the cache memories start to be activated, the data processing unit is connected to the system bus so that data can be transmitted/received between the data processing unit and peripheral devices.