Abstract: The present invention discloses a pipeline processor system comprising a pipeline processor and a memory device, wherein the memory device is comprised of a memory unit for holding data and/or an instruction as well as being accessed to implement memory read operation or memory write operation in a clock cycle; and a data latch unit for latching data to be written into the memory unit, while the pipeline processor is comprised of an instruction detection unit for detecting from fetched instructions a first predetermined instruction which directs the latch of the data as well as a second predetermined instruction which directs write of the data at the data latch means into the memory means; and a latch control unit for controlling to latch operation results of the first predetermined instruction to the data latch unit when the predetermined instruction is detected by the instruction detection unit as well as controlling to write the data at the data latch unit into the memory unit when the second predetermined

Abstract: In a digital signal processor, an arithmetic apparatus capable of performing Viterbi decoding processing at a high speed with minimum addition of hardware and least overhead of memory. Pathmetric value and branchmetric value read out from first and second memories on two paths are simultaneously added by an adder at most significant bits and least significant bits thereof. A comparator compares values of the most significant bits and the least significant bits output from the adder to generate a path select signal indicating the value which is pathmetrically smaller. The select signal is stored in a shift register on a bit-by-bit basis. Of the values of the most significant bits and the least significant bits of a register storing the output of the adder, the smaller one as decided by the path select signal is written in the memory at eight most significant bits or least significant bits thereof via distributor, a bus and a register.