Abstract: A pipelined, microcoded CPU employs conditional branching in microcode execution Data path conditions produced by one microinstruction are used in the selection of a following microinstruction. In high-performance systems, multiple cycle microbranch latency requires that the generation of microbranch conditions be pipelined. Usually a microbranch condition is used exactly once, at the earliest possible time, when dynamic microbranch conditions are only valid a fixed number of microinstructions later in the pipeline. Flexibility of the microcode algorithm is increased by selectively inhibiting the update of the dynamic conditions to delay the use of the condition by one or more cycles, under microcode control, thereby implementing dynamic microbranches, while allowing use of previous dynamic microbranch state.

Abstract: A method of specifying the operands for a microcoded CPU employs a combination of a set of microinstruction routines for generic operand modes, along with hardware primitives for selecting various specific types of operand treatment. Decoding of a machine-level instruction produces an entry point for the microstore, selecting one of the set of generic operand modes. Also, decoding of the instruction produces control bits that are used directly to select the specific operand type or used by the hardware primitives. In this way, branching is avoided in the microinstruction sequences used for operand specifying, but yet the amount of microcode needed is a minimum.

Abstract: A high-performance, pipelined CPU in which an improved method is used for saving registers in memory upon the occurrence of a procedure CALL or RETURN. The registers which need to be saved are defined by a bit-mask, and the number of bits is counted by a hardwired circuit, in each machine cycle, producing an output in the form of an offset from a stack pointer which represents the highest memory location needed to save the registers being used. Then a memory probe can be done to see if this location is writable. Thus, in one microinstruction cycle, the count is made and the memory probe can begin.

Abstract: A method for reading data blocks from main memory by central processing units in a multiprocessor system containing write-back caches. Load or gather instructions contain a write-intent flag. The status of the write-intent flag is determined. It is also determined whether a data block requested in the instruction by one of the processors is located in a corresponding cache, and if so, the requested data block is returned to the processor. If the data block is not in the cache and the write-intent flag indicates that the block will not be modified, the data block is read from main memory without obtaining a write privilege. The requested data block is subsequently returned from the cache to the processor. If the data block is not in the cache and the write-intent flag indicates the data block will be modified by the processor, then the data block is read from main memory while obtaining the write privilege. Subsequently, the requested data block is returned from the cache to the processor.

Abstract: A memory stack used for storing microinstruction addresses in a pipelined CPU is constructed as a last-in, first-out memory using a stack pointer which applies a read control to one location of the stack and applies a write control to the next higher location. An unconditional read and write is done every machine cycle, before a microinstruction could be decoded, then the data on the read bus, or data from the write bus, is used and the pointer is incremented or decremented if a stack Push or Pop is decoded. These correspond to a Call or Return microinstruction. Thus the delay in decoding the microinstruction does not prevent completion of the stack operation in one machine cycle.

Abstract: Bubble compression in a pipelined central processing unit (CPU) of a computer system is provided. A bubble represents a stage in the pipeline that cannot perform any useful work due to the lack of data from an earlier pipeline stage. When a particular pipeline stage has stalled, the CPU instructions that have already passed through the stage continue to move ahead and leave behind vacant stages or bubbles. If a bubble is introduced into a pipeline and the pipeline subsequently stalls, the disclosed CPU takes advantage of this stalled condition to compress the previously introduced bubble.

Abstract: A pipelined CPU employs separate microinstruction pipelines for the execution unit and memory management unit. Deadlocks can occur in a pipelined CPU when there is data dependency in two consecutive instructions. The later instruction may stall the pipeline if operands fetched by an earlier instruction are needed, but the earlier instruction is not producing the memory request for the operands because the pipeline is stalled; this results in a deadlock. Using separate micro-pipelines, the earlier instruction is advanced independently of the rest of the pipeline, in the case of a deadlock, so that the operands for the later instruction are provided and the deadlock is broken.

Abstract: A protocol for transferring instructions between asynchronous processors in a computer system is provided. Each instruction transfer requires the transfer of an opcode and a variable number of operands. The transfer is accomplished via a bus which interconnects the processors. The opcode and operands are assembled in a buffer in the sending processor and then transferred to the receiving processor in reverse order, i.e., operands first and opcode last. The receiving processor does not acknowledge any of the transfers until it receives the opcode which is always sent last. Upon receipt of the opcode, the receiving processor knows the instruction transfer is complete and sends the acknowledge signal immediately thereafter.