摘要:
In a data processing system having an instruction buffer, inconsistency between an instruction stored in a main memory and the same instruction present in the instruction buffer occurs at the time of a store operation to the memory if the instruction supplied to the store area of the main memory previously has been prefetched to and is present in the instruction buffer. The changing in the memory of the content of the instruction prefetched to the instruction buffer and the fact that the changed instruction is about to be executed are detected on the basis of the instruction fetch address of an instruction to be fetched from the memory to the instruction buffer, the instruction address of the next instruction to be executed, information concerning the byte position at which the instruction to be executed is stored, the store address at which the changed instruction is stored in the main storage and information concerning the byte position in the main storage of the instruction which is changed by the store operation. When the execution of the instructions proceeds to the changed instruction, the content of the instruction buffer is invalidated.
摘要:
A gathering system in an electronic computer including detecting means for detecting the occurrences of each of a plurality of events, a counter group having a plurality of counters each of which counts the number of occurrences of corresponding event until the number thereof becomes 2.sup.l, a storage having a plurality of memory areas, each storing 2.sup.K.multidot.l occurrences of the corresponding event and a processing unit for gathering data representative of the number of occurrences of each of the events from the counter and the storage unit.
摘要:
A data processing system, for executing each instruction, by carrying out a plurality of successive partial processing operations, begins to process the first partial processing operation of an instruction succeeding a defeat overlap instruction before execution of the defeat overlap instruction is finished in response to the detection that a predetermined number of machine cycles is further required for the complete execution of the defeat overlap instruction.