Digital Equipment Corporation

Digital Equipment Company

Douglas D Williams

Richard B Gillett, Jr

David Martin Fenwick

Timothy J Stanley

Darrel D Donaldson

Richard T Witek

Rebecca L Stamm

Douglas J Burns

Richard Heye

John M Parchem

David A Orbits

David M Robinson

Mark N Howard

A load/store pipeline in a computer processor for loading data to registers and storing data from the registers has a cache memory within the pipeline for storing data. The pipeline includes buffers which support multiple outstanding read request misses. Data from out of the pipeline is obtained independently of the operation of the pipeline, this data corresponding to the request misses. The cache memory can then be filled with the requested for data. The provision of a cache memory within the pipeline, and the buffers for supporting the cache memory, speed up loading operations for the computer processor.

Pipeline utilizing an integral cache for transferring data to and from a

An interrupting node for providing interrupt requests to a pended bus. The interrupting node provides to the pended bus an interrupt request message including ID data for identifying the interrupting node as the source of an interrupt request. An interrupt servicing node provides interrupt acknowledge messages including destination data specifying a particular interrupting node at times when the interrupt servicing node is ready to service an interrupt request message. The interrupting node detects whether an interrupt acknowledge message on the bus includes destination data specifying that interrupting node, and provides an interrupt vector message to the bus in response.

Interrupting node for providing interrupt requests to a pended bus

A multiple node computer system includes processor nodes, memory nodes, and input/output nodes interconnected on a pended bus. The system includes a lockout indicator which is set upon receipt of a locked response message by a processor node from a memory node in response to an interlock read command. The processors include a lockout check circuit responsive to the condition of the lockout indicator and will restrict generation of additional interlock read commands according to a predetermined access gating criterion until the lockout indicator is reset. In this manner, processor nodes of the system are assured equitable access to a memory node.

Commander node method and apparatus for assuring adequate access to

A prediction logic device operating in conjunction with a vector processor to predict, before the completion of the translation of the virtual addresses of all of the data elements of a vector, the valid performance of all virtual-address to physical-address translations for the data elements of the vector. The prediction logic device asserts an MMOK signal to a scalar processor when it becomes known that no memory management fault and/or translation buffer miss will occur such that the scalar processor can resume vector instruction issue to the vector processor at the earliest possible time.

System for predicting memory fault in vector processor by sensing

A cache coherency protocol for a multi-processor system which provides for read/write, read-only and transitional data states and for an indication of these states to be stored in a memory directory in main memory. The transitional data state occurs when a processor requests from main memory a data block in another processor's cache and the request is pending completion. All subsequent read requests for the data block during the pendency of the first request are inhibited until completion of the first request. Also provided in the memory directory for each data block is a field for identifying the processor which owns the data block in question. Data block ownership information is used to determine where requested owned data is located.

Cache coherency protocol for multi processor computer system

Method and apparatus for predicting valid performance of virtual-address

A load/store pipeline in a computer processor for loading data to registers and storing data from the registers has a cache memory within the pipeline for storing data. The pipeline includes buffers which support multiple outstanding read request misses. Data from out of the pipeline is obtained independently of the operation of the pipeline, this data corresponding to the request misses. The cache memory can then be filled with the data that has been requested. The provision of a cache memory within the pipeline, and the buffers for supporting the cache memory, speed up loading operations for the computer processor.

Pipeline having an integral cache which processes cache misses and loads

Apparatus and method for servicing interrupt requests on a pended bus. An interrupt servicing node provides interrupt acknowledge messages including destination data specifying a particular interrupting node at times when the interrupt servicing node is ready to service an interrupt request message. The interrupt servicing node includes storage for specifying the identity of a particular interrupting node and for indicating that an interrupt request is pending from a particular interrupting node. An interrupting node provides to the pended bus an interrupt request message including ID data for identifying the interrupting node as the source of an interrupt request. The interrupting node detects whether an interrupt acknowledge message on the bus includes destination data specifying that interrupting node.

Apparatus and method for servicing interrupts utilizing a pended bus

A computer system having multiple processors interconnected by a pended bus provides exclusive read-modify-write operations employing multiple lock bits. A processor generates an interlock read command which is transmitted as a transfer over the pended bus to a memory or I/O node. Acknowledge confirmations are transmitted by the memory back to the processor two bus cycles after each bus cycle of the processor transfer. The processor transfer, including an interlock read command, is stored in a input queue in memory and processes in turn by the memory. A first interlock read command to a specified memory location causes a lock bit to be set for that location and a first type of response message including the contents of the specified location to be generated by the memory and stored in an output queue. The memory obtains access to the pended bus through an arbitration process and transmits a response message including the contents of the memory location specified in the interlock read command at an unspecified time after initiation of the interlock read command. A subsequent interlock read command from the processor to the same memory location will result in a denial of access to the specified location and in the generation of a second type of response message by the memory which indicates that the specified location is locked.

Method and apparatus for managing multiple lock indicators in a

Interrupt servicing node for servicing interrupt requests on a pended bus. The interrupt servicing node provides interrupt acknowledge messages including destination data specifying a particular interrupting node at times when the interrupt servicing node is ready to service the interrupt request message. The interrupt servicing node includes storage elements for indicating whether an interrupt request is pending from a particular interrupting node. An interrupt request message on the bus includes ID data for identifying a particular interrupting node as the source of an interrupt request.

Node for servicing interrupt request messages on a pended bus

A memory node in a computer system having multiple processors interconnected by a pended bus provides exclusive read-modify-write operations employing multiple lock bits. A processor generates an interlock read command which is transmitted as a transfer over the pended bus to the memory node. Acknowledge confirmations are transmitted by the memory back to the processor two bus cycles after each bus cycle of the processor transfer. The processor transfer, including an interlock read command, is stored in a input queue in memory and processed in turn by the memory. A first interlock read command to a specified memory location causes a lock bit to be set for that location and a first type of response message including the contents of the specified location to be generated by the memory and stored in an output queue. The memory obtains access to the pended bus through an arbitration process and transmits a response message including the contents of the memory location specified in the interlock read command at an unspecified time after initiation of the interlock read command. A subsequent interlock read command from the processor to the same memory location will result in a denial of access to the specified location and in the generation of a second type of response message by the memory which indicates that the specified location is locked.

System for implementing multiple lock indicators on synchronous pended

A processor node providing exclusive read-modify-write operations in a computer system having multiple processors interconnected by a pended bus and employing multiple lock bits. The processor generates an interlock read command which is transmitted as a transfer over the pended bus to a memory or I/O mode. Acknowledge confirmations are transmitted by the memory back to the processor two bus cycles after each bus cycle of the command transfer. The command transfer, including an interlock read command, is stored in an input queue in memory and is processed in turn by the memory. A first interlock read command to a specified memory location causes a lock bit to be set for that location and a first type of response message including the contents of the specified location to be generated by the memory and stored in an output queue. The memory obtains access to the pended bus through an arbitration process and transmits a response message including the contents of the memory location specified in the interlock read command at an unspecified time after initiation of the interlock read command. A subsequent interlock read command from the processor to the same memory location will result in a denial of access to the specified location and in the generation of a second type of response message by the memory which indicates that the specified location is locked.

Method and apparatus for initiating interlock read transactions on a

Method and apparatus for assuring adequate access to system resources by

An interface system between a high speed user bus and a system bus is provided to present to the user bus a picture of the data transferred on the system bus every clock cycle of that system bus. The interface system also allows the user bus to transfer data back to the system bus during selected bus cycles. By using a single pin connection to the system bus, the user bus can send communications back to itself by way of the system bus.

Growing community of inventors

Pepperell, MA, United States of America

Douglas D Williams