An improvement is disclosed for a computer system wherein a high speed peripheral device sends data words of a first length to a memory which operates with a word of a greater length, for example, twice the data word length. The improvement includes an apparatus for identifying pairs of data words which belong together in a single memory word. Then, a pair of the data words can be written to memory in a single operation, rather than two separate write operations.
Apparatus is disclosed for buffering writes from a CPU to main memory, in which sequential write requests to the same address are gathered and combined into a single write request. The embodiment described does not permit gathering with the write request in the buffer which is next scheduled for action by the main memory bus controller, nor does it permit gathering with other than the immediately preceding write request. The invention is implemented using a plurality of buffer ranks, each comprising a data rank, an address rank, and a valid rank for indicating which bits or bytes of the data rank contain data to be written to memory.
A memory address pointer that selects a memory location that is mapped to a video graphics circuit port is incremented only when all bytes in a memory location have been read from or written to by the host CPU. This does not depend on the order in which the host CPU reads or writes data bytes. Therefore a video controller that uses the present invention will work with 8 bit, 16 bit as well as high performance 32 bit input/output instructions.
A pipelined register cache for increasing a computer processor's execution speed by reducing the time required to access register data. A register cache is implemented to keep often-used registers in high-speed storage immediately available to the processor's arithmetic and logic unit (ALU). The register cache is constructed using a number of individual register stages which are connected in series such that the register information contained in each register stage is passed from one register stage to the next in a First-In, First-Out (FIFO) queue arrangement. Each register stage stores a register address tag for identifying the particular primary register being represented in that register stage, and a data value representing the actual register contents. When a register that is not represented in the cache is needed for a calculation, the register information is first loaded from the primary register storage into the first register stage of the register cache. Once the register is represented in register cache, it can be accessed quickly by the arithmetic and logic unit for computations. As new register information is loaded into the register cache, the older register information in each register stage is passed from one register stage to the next. When register information overflows from the last register stage it is then placed back into the primary register storage.
A method, apparatus and system for storing data wherein pieces of data are selectively combined at times when a storage device is not able to accept requests to store data for a duration sufficient that multiple additional pieces of data have been received since the last request to store data was made and at least two of those pieces of data are able to be combined into a piece of data that is better configured for storage.
A register selection mechanism for an instruction prefetch buffer which allows instructions having different lengths to be accessed on the instruction boundaries. The instruction prefetch buffer comprises a one-port-write, two-port-read array (10). Address generation and control logic (16) is responsive to a read pointer (15) for controlling access to odd and oven addresses in the array. Additional logic may be provided to provide an indication that the instruction prefetch buffer is empty.
