Digital Equipment Corporation

Hewlett-packard Development Company, L.p.

Raj K Ramanujan

Simon C Steely, Jr

James B Keller

David J Sager

Peter J Bannon

Paul J Lemmon

Jay C Stickney

Robert E Stewart

William A Samaras

John DeRosa

John H Zurawski

Steven Ho

Walter A Beach

Anne S Valiton

John De Rosa

Jeffrey L Bell

An apparatus comprises a plurality of ports wherein each port is adapted to couple to a device. At least one port connects by way of first and second unidirectional, point-to-point communication links with a device. The first unidirectional, point-to-point communication link transfers data from the device to the central logic unit and the second unidirectional, point-to-point communication link transfers data from the central logic unit to the device.

High speed bus system that incorporates uni-directional point-to-point buses

A high speed bus system for use in a shared memory system that allows for the high speed transmissions of commands and data between a number of processors and a memory array of a multi-processor, shared memory system, with the high speed bus system including a central unit and a series of uni-directional buses that connect between the plurality of processors and shared memory, with the central unit including arbitration logic and a series of multiplexers to determine which CPUs are granted access to shared buses, scheduling logic that works with the arbitration logic and multiplexers to determine which CPUs are granted access to the shared buses, and port logic for combining the CPU transmissions and determining if such transmissions are valid.

A computer system is adapted to transfer write data from a central processing unit to one of a plurality of memory modules in a memory array by transferring a block of write data to a memory control logic device. The memory control logic device transfers the block of data in a plurality of data bursts interspaced by a preselected number of bus cycles. During the interspaced preselected number of bus cycles, the memory control logic device sends pending read commands to an available memory module thereby overlapping read and write operations on the memory bus, thus, lowering memory read latency.

Interleaving read and write operations on a bus and minimizing buffering on a memory module in a computer system

A lockout avoidance circuit is provided for a plurality of nodes which generate lock requests for a shared resource such as a memory location. The circuit insures that lock requests are eventually satisfied. A lock queue includes a plurality of registers pipelined together. Lock requests only enter the lock queue if they are refused access to a shared resource a predetermined number of times. A first register is the head of the queue and the last register is the bottom of the queue. An enabling circuit allows the queue to store in the registers lock requests received from the different nodes in the order in which they are initially refused service. The enabling circuit operates the queue by pushing the stored lock requests toward the head of the queue each time the head entry in the queue is serviced. The lockout avoidance circuit is implemented at each level of the system wherein a lockout condition can occur.

Apparatus and method for ensuring that lock requests are serviced in a

The bandwidth of a first bus and a second bus, unequal due to differences in protocol overheads and cycle times between the buses, are equalized without sacrificing any bandwidth on the lower bandwidth bus and without introducing any buffering in a control logic device. The control logic device equalizes the bandwidths of the buses by instructing a device coupled to the second bus to insert a partial dead bus cycle in a read transmission thereby dynamically adjusting read timing on the second bus when the second bus is heavily loaded.

Method and apparatus for the dynamic adjustment of data transfer timing

A cache coherency scheme in a multiprocessor computer system allows data sharing between caches at a fast rate. A new cache coherency state is introduced which allows a processor pair to more effectively share data and eliminate bus transfers thereby improving system throughput. The transfer of data is accomplished by the returning a portion of a preselected data block pursuant to either a read or a read for ownership request by a first one of the processors of the processor pair by the second processor of the processor pair. The ownership of the portion of the preselected data block is shared by the processor pair. Both processors set an indicator to denote that the preselected data block is an incomplete data block.

Method and apparatus for sharing data between processors in a computer

A method and apparatus for optimizing the performance of a multiple cache system computer having separate caches for data and instructions in which all writes to the data cache are monitored. If the address tag of the item being written matches one of a list of tags representing valid instructions currently stored in the instruction cache, a flag called I.sub.-- FLUSH.sub.-- ON.sub.-- REI is set. Until this flag is set, REI (Return from Exception or Interrupt) instructions will not flush the instruction cache. When the flag is set, an REI command will also flush or clear the instruction cache. Thus, the instruction cache is only flushed when an address referenced by an instruction is modified, so as to reduce the number of times the cache is flushed and optimize the computer's speed of operation.

System for flushing instruction-cache only when instruction-cache

A system and method for controlling a shared memory bus in a computer of a multi-processor system prevents collisions on the shared bus and ensures that the bus is full at system start-up. Steady state operations are maintained without the need for a queuing mechanism in the system's memory controller and in view of the memory modules of the shared memory having different read access times, with the system and method being implemented in a system that includes a central unit and multiple uni-directional buses that are disposed between a shared memory and a plurality of processors, with the central unit controlling access to, and use of, the shared buses of the system.

Method of controlling a shared memory bus in a multiprocessor system for

During the operation of a computer system whose processor is supported by virtual cache memory, the cache must be cleared and refilled to allow the replacement of old data with more current data. The cache is filled with either P or N (N&gt;P) blocks of data. Numerous methods for dynamically selecting N or P blocks of data are possible. For instance, immediately after the cache has been flushed, the miss is refilled with N blocks, moving data to the cache at high speed. Once the cache is mostly full, the miss tends to be refilled with P blocks. This maintains the currency of the data in the cache, while simultaneously avoiding writing-over of data already in the cache. The invention is useful in a multi-user/multi-tasking system where the program being run changes frequently, necessitating flushing and clearing the cache frequently.

Cache with at least two fill rates

The invention is directed to a cache memory system in a data processor including a virtual cache memory, a physical cache memory, a virtual to physical translation buffer, a physical to virtual backmap, an Old-PA pointer and a lockout register. The backmap implements invalidates by clearing the valid flags in virtual cache memory. The Old-PA pointer indicates the backmap entry to be invalidated after a reference misses in the virtual cache. The physical address for data written to virtual cache memory is entered to Old-PA pointer by the translation buffer. The lockout register arrests all references to data which may have synonyms in virtual cache memory. The backmap is also used to invalidate any synonyms.

Cache memory system

The invention is directed to a method by which data from a first synchronous subsystem is transmitted to a second synchronous subsystem physically spaced from the first subsystem such that the system clock is skewed relative to the two subsystems. Pursuant to the invention, the transmitted data cycle time is multiple of the common system cycle time and a clock signal is forwarded with the data from the first subsystem to the second subsystem. The clock signal forwarded with the data provides an indication to the second subsystem of the instants of time at which received data changes. The data change instants are used to generate a binary signal that inverts its value at each such instant. The binary signal is transmitted through a synchronizer to produce a logic signal in the second subsystem. The logic signal is utilized to transmit the received data into the second subsystem in synchronization with the clock of the second subsystem.

Method and apparatus for high speed data transmission between two

A cache memory system in a data processor that has a main memory and a processing unit, the cache memory system including a virtually addressed storage cache. This virtually addressed storage cache is connected to the main memory for storing in storage cache locations preselected portions of data from the main memory. Each cache location includes a valid indicator to indicate the data in the cache location is current. A translation buffers is coupled to the storage cache, and translates a virtual address to a physical address. The backmap is coupled to the storage cache and the translation buffer, and invalidates data in the storage cache by generating an invalidate index to the cache location at which a valid indicator is to be cleared only when data in the storage cache is to be invalidated. The cache memory system includes at least one lockout register for storing addresses for data which may exist in more than one storage location. The backmap invalidates all copies of the data in the storage cache after every reference to data in the storage cache using an address in the lockout register.

Growing community of inventors

Leominster, MA, United States of America

Raj K Ramanujan