International Business Machines Corporation

Derrick LeRoy Garmire

Jay Robert Herring

Christine M Desnoyers

Francis Alfred Kampf

Robert Frederick Stucke

Kathy Sue Barkey

Harold Edgar Roman

Daniel Gerard Smyth

Carl Alfred Bender

Scot H Rider

Craig Brian Stunkel

Daniel Francis Casper

Frank David Ferraiolo

Robert Francis Bartfai

Marten J Halma

Robert S Capowski

Donald G Grice

Kevin J Reilly

Nicholas P Rash

Ronald A Linton

Antoinette Elaine Herrmann

Peter Kenneth Szwed

Michael Patrick Muhlada

Donna C Myers

William R Milani

Tim Zhang

Antoinette E Vallone

Vadim M Tsinker

Sheryl M Genco

A method is provided for packet flow control for a switching node of a data transfer network. The method includes actively managing space allocations in a central queue of a switching node allotted to the ports of the switching node based on the amount of unused space currently available in the central queue. In a further aspect, the method includes separately tracking unallocated space and vacated allocated space, which had been used to buffer packets received by the ports but were vacated since a previous management update due to a packet being removed from the central queue. Each port is offered vacated space that is currently allocated to that port and a quantity of the currently unallocated space in the central queue to distribute to one or more virtual lanes of the port.

Method for actively managing central queue buffer allocation

Method, system and program product are provided for packet flow control for a switching node of a data transfer network. The method includes actively managing space allocations in a central queue of a switching node allotted to the ports of the switching node based on the amount of unused space currently available in the central queue. In a further aspect, the method includes separately tracking unallocated space and vacated allocated space, which had been used to buffer packets received by the ports but were vacated since a previous management update due to a packet being removed from the central queue. Each port is offered vacated space that is currently allocated to that port and a quantity of the currently unallocated space in the central queue to distribute to one or more virtual lanes of the port.

System and program product for actively managing central queue buffer allocation

A method is provided for selecting a data source for transmission on one of several logical (virtual) lanes embodied in a single physical connection. Lanes are assigned to either a high priority class or to a low priority class. One of six conditions is employed to determine when re-arbitration of lane priorities is desired. When this occurs a next source for transmission is selected based on a the specification of the maximum number of high priority packets that can be sent after a lower priority transmission has been interrupted. Alternatively, a next source for transmission is selected based on a the specification of the maximum number of high priority packets that can be sent while a lower priority packet is waiting. If initialized correctly, the arbiter keeps all of the packets of a high priority packet contiguous, while allowing lower priority packets to be interrupted by the higher priority packets, but not to the point of starvation of the lower priority packets.

Priority arbitration mechanism

A method and system are provided for processing data packets at a data-transfer network node. The method and system include determining a length of time that a packet has been buffered at the node by associating a timer with each data packet received and buffered in the node's central queue. The central queue subsequently reads the associated timer to determine a length of time that a data packet has been buffered prior to the data packet being transmitted by the node. If a packet has been buffered too long, then the queue discards the packet. Otherwise, the queue permits transmission of the packet. The amount of circuitry in the switching node's central queue is reduced by locating the packet timers in timer logic external to the queue.

Packet processing system and method for a data transfer node with time-limited packet buffering in a central queue

A massively parallel system has a self-timed interface (STI) in which a clock signal clocks bit serial data onto a parallel, electrically conductive bus and the clock signal is transmitted on a separate line of the bus. The received data on each line of the bus is individually phase aligned with the clock signal. Digital data is transmitted at high speeds via the parallel bus to provide a scalable communications network for parallel processing systems while eliminating precise bus length and system clock rates as a critical or limiting factor.

Synchronous interface for transmitting data in a system of massively parallel processors

Received portion of message is stored persistently and transmitted without awaiting receipt of another portion of the message and without generating a new message. The storing and transmitting can occur substantially simultaneously and be performed by one or more hardware elements. Originator of the message can choose whether to indicate indication of broadcasting of the message. First hardware element can determine local acknowledgement for message and second hardware element can determine determinative signal of the local acknowledgement and at least one of: one or more collected intended recipient acknowledgements for the message; and one or more collected determinative signals of intended recipient acknowledgements for the message.

Parallel computer network broadcasting and acknowledgement

A credit-based flow control checking scheme is presented for controlling data communications in a closed loop system comprising a sender, a receiver and a link coupling the sender and receiver. The credit-based scheme includes automatically periodically transmitting a credit query from the receiver to the sender and upon return receipt of a credit acknowledge containing the available credit count maintained by the sender, determining whether credit gain or credit loss has occurred subsequent to initialization of the closed loop system. Along with automatically determining whether credit gain or credit loss has occurred, a method/system is presented for automatically correcting the loss or gain without requiring resetting of the closed loop system.

Credit-based flow control checking and correction method

A quiesced and synchronous distributed data processing system includes a primary node, a secondary node and a switch between the primary and secondary nodes. The TOD for the primary node is set depending on the system. The invention sets the TOD for the switch based on the TOD of the primary node. The primary node then verifies the switch TOD. If verified, the TOD for the secondary node is set, based on the TOD of the primary node. Finally, the secondary node self-verifies its TOD.

Verifying a time-of-day counter

A self-timed link between two elements in a computer system is initialized. Each element sends an initialization request to the other. If successfully received, the elements exchange signals with oscillation-free segments over multiple clock cycles. If successful, the elements indicate to each other that initialization is complete. Optionally, a link operation parameter can be sent with the initialization complete indication for post-initialization link control.

Method for initializing a self-timed link

An error message is generated by a hardware element of a distributed computer system, when an error is detected. The error message is then forwarded from the hardware element to one or more designated processing nodes of the distributed computer system. The hardware element includes, for instance, a switch element or a communications adapter adapted to report detected errors.

Growing community of inventors

Kingston, NY, United States of America

Derrick LeRoy Garmire