A method for detecting and handling memory-mapped I/O in a pipelined data processing system is provided. The method uses two signals on the system interface: when the system generates a read bus cycle, it activates an output signal if certain I/O requirements are not satisfied; an input signal is activated when the reference is to a peripheral device that exhibits certain characteristics; when the system detects that both the input signal and the output signal are active, it discards the data read during the bus cycle, serializes instruction execution and regenerates the read bus cycle, this time satisfying the requirements for I/O such that the output signal is driven inactive.

The present invention relates to an access request control apparatus and more specifically to an apparatus for determining priority between a plurality of access requests in a memory control apparatus which uses a pipeline. One of the access requests from a plurality of channel processing devices CHP's is selected by a first priority determination circuit. The selected CHP request, the requests from a plurality of central processing units and the request in the loop-back of the pipeline control circuit are considered for selection by a second priority determination circuit. In case a CHP request, selected by the first priority determination circuit, is not selected by the second priority determination circuit or selected but nullified in the course of the pipeline, the CHP request is returned to the first priority determination circuit. But, in this case, a higher priority is given to the CHP request in the first priority determination circuit. In addition, the priority algorithm in the second priority determination circuit considers the kinds of operations of each access request and highly efficient memory access control can be realized.

A memory system includes a memory, an input circuit and a logic circuit. The input circuit is coupled to receive a memory address and, during a write operation, the corresponding write data to be written into the SRAM. The logic circuit causes the write data to be stored in the input circuit for the duration of any sequential read operations immediately following the write operation and then to be read into the memory during subsequent write operation. During the read operation, data which is stored in the write data storage registers prior to being read into the memory can be read out from the memory system should the address of one or more read operations equal the address of the data to be written into the memory while temporarily stored in the write data storage registers. Thus, no "bus turnaround" down time is experienced by the system thereby increasing the bandwidth of the system. The system can operate in a single pipeline mode or a dual pipeline mode.

A synchronous memory circuit is capable of double data transfer rate per clock cycle, 100% bus utilization (i.e., no idle clock cycles in bus turn arounds), and has only one clock cycle of latency in each of read and write burst operations. The memory circuit has a data bus 202, at least two memory blocks (20, 30), a multiplexer (120) for receiving two read data items from respective two memory blocks in a read burst operation and allowing one of the two read data items to be provided on the data bus half a clock cycle after the read burst operation is initiated and allowing the other one of the two read data items to be provided on the data bus one clock cycle after the read burst operation is initiated, and two registers (50, 70) for storing respective two write data items provided on the data bus in a first write burst operation wherein one of the two write data items is written to one of the two memory blocks at the initiation of a next write burst operation following the first write burst operation and the other one of the two write data items is written to the other one of the two memory blocks half a clock cycle after the initiation of the next write burst operation, wherein in two consecutive clock cycles the two write data items are capable of being transferred to the memory circuit via the data bus and the two read data items are capable of being transferred from the memory circuit via the data bus.

A memory system including a memory array, an input circuit and a logic circuit is presented. The input circuit is coupled to receive a memory address and a set of individual write controls for each byte of data word. During a write operation, the input circuit also receives the corresponding write data to be written into the SRAM. The logic circuit causes the write data and write control information to be stored in the input circuit for the duration of any sequential read operations immediately following the write operation and then to be read into memory during a subsequent write operation. During the read operation, data which is stored in the write data storage registers prior to being read into the memory can be read out from the memory system should the address of one or more read operations equal the address of the data to be written into the memory while temporarily stored in the write data storage registers. The logic circuit also detects which bytes of data are not to be written into the SRAM so that, during a read operation, those bytes not to be written into the SRAM are read from the SRAM in order to output a complete word corresponding to the value at the read address. No "bus turnaround" down time is experienced by the system thereby increasing the bandwidth of the system. The system can operate in a single pipeline mode or a dual pipeline mode.