International Business Machines Corporation

Gregory F Grohoski

Oscar R Mitchell

James Allan Kahle

Paul Joseph Jordan

David Scott Ray

Charles Roberts Moore

John Cocke

Randall Dean Groves

Ravindra K Nair

Steven G Ludwig

Myhong Nguyenphu

William R Hardell, Jr

Vojin G Oklobdzija

Chien-Chyun Lee

Yongjae Rim

Tung M Nguyen

Tung Manh Nguyen

Yonjae Rim

John F Kearns

Edmond J Boufarah

This invention relates to a system and method for adapting the ports of computing elements in transferring data between computing elements on a network. Ports of the computing elements are interconnected for data transfer through a switch complex. The interconnected ports are adapted to cooperate together in transferring the data. Data which is ordinarily designated to be transferred through one port of a computing element may be transferred through a different port.

System and method for controlling the direction of data flow between

A small buffer called a branch-anticipate buffer (BAB) is used to store groups of instructions which are likely to be required from the instruction cache (I-cache) when an instruction prefetch miss occurs. When a prefetch miss occurs, the BAB is checked to see if instructions corresponding to the target address are available. If they are available, these instructions are copied into an appropriate buffer. If the instructions corresponding to the target address are unavailable, these instructions are fetched from the I-cache and placed into a buffer and, selectively, into the BAB. The BAB only maintains branch target addresses that have not been previously scanned and that cannot be prefetched in time. This allows for smaller buffer sizes, and resulting quicker access time, when checking the BAB for instructions to be executed by a processor.

Method and apparatus for managing the execution of instructons with

A computer networking system includes a cross bar switch and a protocol for operating the same. The crossbar switch typically connects a plurality of ports one to another and the protocol establishes a connection between a first desired port and a second desired port selected from the plurality of ports. Each port further connects to a compute element via a master bidirectional bus and a slave bidirectional bus. Any of the compute elements can serve as either a master or slave to any other compute element connected to the crossbar switch. A master port connects the bidirectional bus to the crossbar switch and a slave port connects the slave bidirectional bus to the crossbar switch. The master port is reserved for compute element initiated operations while the slave port is reserved for network initiated operations. The crossbar switch receives and transmits control, address, and data information over a CPU bus to a network router unit, which is used as the interface between the CPU and the crossbar switch to translate CPU bus protocol to switch protocol. The crossbar switch uses an in band switch design, which excludes explicit control pins to control the switch and relies on control being transmitted over wires shared between the data and the address functions. The switch and protocol support load, store, broadcast, compare, and swap and barrier synchronization primitives, using no control pins and minimizing overhead.

Crossbar switch apparatus and protocol

A data processing system including a circuit for storing a plurality of instructions in a sequence together with a circuit for fetching a plurality of instructions. A circuit is provided for dispatching a plurality of the instructions to one or more processors for execution during a single computation cycle. A control circuit is connected to the dispatching circuit to delay the dispatching of an instruction. when the instruction has an execution result that is dependent upon a previous instruction execution that will set at least one bit in a condition register. The delayed instruction is delayed until that condition register has been accordingly set.

Single cycle dispatch delay in a multiple instruction dispatch mechanism

A data processing system for executing a sequence of instructions. The data processing system includes several processors each for executing instructions. Also included is a dispatching apparatus for dispatching each of the instructions to one of the processors. Control circuitry is included for directing the concurrent execution of the dispatched instructions in the processors irrespective of the location of the instructions in the sequence. The control circuitry includes the capability to receive an instruction interrupt signal. The control circuitry then determines which instruction generated the instruction interrupt. Upon this determination, the control circuitry resets the processors and the dispatching apparatus to the state that existed when the instruction that generated the instruction interrupt was earlier executed in order to re-execute the instruction that caused the interrupt signal in accordance with its location in the instruction sequence.

Parallel processor instruction dispatch apparatus with interrupt handler

A data processing system including a circuit for storing a sequence of instructions, a circuit for determining if the instruction sequence includes a branch instruction, a circuit for storing a sequence of branch target instructions in response to the determination of the existence of a branch instruction in the stored sequence of instructions, a circuit for dispatching instructions in sequence after the branch instruction to a processor to be executed on condition that a branch is to be taken before a determination of whether said branch will be taken and simultaneously for determining if the branch is to be taken, any circuit for directing the processor to execute the instructions in sequence after the branch if the branch is not taken, or, if the branch is to be taken, for dispatching the branch target instruction sequence to the processor for execution.

System for reducing delay in instruction execution by executing branch

A reduction in the number of cycles required to obtain data from main storage when a 'miss' occurs in a cache for a desired line of data but a match to another line from the same page of data in main storage as the desired line is present in the cache. In accordance with the present invention if a match to another line from the same page is present, the real address for the other line from the same page is used to fetch the desired line of data directly from the main storage without an address translation. This technique works for a virtually addressed cache whose directory contains both a virtual and a real address for every line of data stored in the cache.

Cache performance in an information handling system employing page

A data processing system including an instruction storage buffer for storing a sequence of instructions requiring an operation by at least two processors. The two processors are provided that execute instructions from the instruction storage buffer. An instruction dispatch circuit is provided that dispatches the instructions to the processors. At least one processor includes the capability to execute dispatched instructions before the execution of a preceding instruction in the instruction sequence by another processor. Also, at least one processor includes the capability to delay execution of an interruptable instruction until the instruction can be executed in its appropriate sequential order in the sequence. Also, upon the occurrence of the interrupt, the processors include the capability to purge the instruction storage buffer in order that the interrupt software instructions may be stored for execution.

Tightly coupled multiprocessor instruction synchronization

A floating point instruction control mechanism which processes loads and stores in parallel with arithmetic instructions. This results from register renaming, which removes output dependencies in the instruction control mechanism and allows computations aliased to the same register to proceed in parallel.

Growing community of inventors

Cedar Park, TX, United States of America

Gregory F Grohoski