International Business Machines Corporation

Paul Allen Ganfield

David John Krolak

Tolga Ozguner

Kent Harold Haselhorst

Mark David Bellows

William John Starke

Lonny Lambrecht

Jeffrey A Stuecheli

Leland Leslie Day

Melissa Ann Barnum

Guy L Guthrie

Michael Steven Siegel

Charles Francis Marino

Ryan Abel Heckendorf

Steven Michael Douskey

John David Irish

Kerry Christopher Imming

Christian Jacobi

Derek E Williams

Luis Alfonso Lastras-Montano

Patrick James Meaney

Anthony Gus Aipperspach

Praveen S Reddy

Glenn David Gilda

Jentje Leenstra

Charles Luther Johnson

John Thomas Hollaway, Jr

Ashutosh Mishra

Logan I Friedman

Charles Ray Johns

Daniel Frank Moertl

Todd Alan Christensen

James David Strom

Peichun Peter Liu

Michael B Spear

Daniel Guy Young

Jason Andrew Thompson

Roopesh Ambalath Matayambath

Guy Richard Currier

Dirk Allmendinger

Brian M McKevett

Yvonne Hanson Kleppel

James Maurice Wallin

Paul W Wong

Ryan Abel Heakendorf

Michael Frank Carnevale

Murali Vaddigiri

Sven Boekholt

Embodiments of the invention are directed to a computer-implemented method of operating a data transmission system. The data transmission system includes a transmitter and a receiver. The computer-implemented method includes using the transmitter to send serialized data from the transmitter through a plurality of lanes to the receiver. The transmitter sends the serialized data at a first serialization ratio. The receiver is configured to receive and load the serialized data at a second deserialization ratio, wherein the first serialization ration is greater than the second deserialization ratio.

Low-latency deserializer having fine granularity and defective-lane compensation

A computer-implemented method includes using a transmitter to send data from the transmitter through a plurality of lanes to a receiver using a synchronous operation mode that includes sending the data from the transmitter through the plurality of lanes to the receiver in a synchronous transmission manner that relies on an alignment between a transmitter clock frequency and a receiver clock frequency. A synchronous operation performance analysis (SOPA) is performed during the synchronous operation mode. A switch from the synchronous operation mode to an asynchronous operation mode is made based on a result of performing the SOPA. The asynchronous operation mode includes sending the data from the transmitter through the plurality of lanes to the receiver without requiring alignment between the transmitter clock frequency and the receiver clock frequency.

Low-latency, high-availability and high-speed SerDes interface having multiple synchronization modes

A time of day (TOD) synchronizer in a first processor transmits a latency measure message simultaneously on two links to a second processor. In response, the receiver in the second processor detects latency differential between the two links, detects the delay in the second processor, and sends the latency differential and delay to the first processor on one of the two links. The first processor stores TOD delay values in the two links that account for the latency differential between the two links. When a TOD message needs to be sent, a link loads a counter with its stored TOD delay value, then decrements the counter until the TOD message is ready to be sent. The resulting counter value is the receiver delay value, which is transmitted to the receiver as data in the TOD message, thereby reducing TOD jitter between the two links.

Reducing time of day latency variation in a multi processor system

In a serial communication interface with transceivers that run on different clocks, an ACK transmit FIFO is used to track packets transmitted, and an ACK receive queue is used to track ACK bits for received packets. The ACK receive queue contains a number of entries, and training for the transceivers begins transmitting ACK bits from the ACK receive queue once the ACK receive queue has multiple valid ACK bits. When the ACK receive queue is less than a lower threshold, an ACK compensation mechanism sends one or more packets that make the ACK receive queue grow. When the ACK receive queue is more than an upper threshold, the ACK compensation mechanism sends one or more packets that make the ACK receive queue shrink. The combination of the ACK receive queue and the ACK compensation mechanism allow dynamically compensating for the different clocks of the two transceivers.

ACK clock compensation for high-speed serial communication interfaces

Data processing in a data processing system including a plurality of processing nodes coupled to an interconnect includes receiving, by a fabric controller, a first command from a remote processing node via the interconnect. The fabric controller determines that the command includes a replay indication, the replay indication indicative of a replay event at one or more processing nodes of the plurality of processing nodes. The first command is dropped from a deskew buffer of the fabric controller responsive to the determining that the command includes the replay indication.

Link-level cyclic redundancy check replay for non-blocking coherence flow

Data processing in a data processing system including a plurality of processing nodes coupled by a communication link includes receiving a first command from a first processing node. A link stall of the communication link is detected by a first link layer of the first processing node. A stop command is received at a first transaction layer of the first processing node from the first link layer. The first command is truncated by the first transaction layer into a first truncated command responsive to receiving the stop command. A command arbiter is instructed to stop issuing new commands. The first truncated command is forwarded to an asynchronous crossing buffer of the first processing node.

Dropped command truncation for efficient queue utilization in multiprocessor data processing system

A time of day (TOD) synchronization mechanism in a first processor transmits a latency measure message simultaneously on two links to a second processor. In response, the receiver in the second processor detects latency differential between the two links, detects the delay in the second processor, and sends the latency differential and delay to the first processor on one of the two links. The first processor stores TOD delay values in the two links that account for the latency differential between the two links. When a TOD message needs to be sent, a link loads a counter with its stored TOD delay value, then decrements the counter until the TOD message is ready to be sent. The resulting counter value is the receiver delay value, which is transmitted to the receiver as data in the TOD message, thereby reducing TOD jitter between the two links.

Reducing time of day latency variation in a multi-processor system

A method, data processing system, and computer program product for verifying the functionality of a digital circuit. The method includes transmitting sequences of parallel data packets via parallel data transfer paths. Prior to receipt of at least two of the transmitted sequences, a first skew is introduced between the at least two of the transmitted sequences. This introduction includes inserting one or more parallel control data packets in the transmitted sequences, and erasing one of the control data packets, replacing one of the control data packets, and inserting another control data packet in one of the sequences. The method includes determining if an expected indicator signal is provided in the form of an overflow indicator or an underflow indicator.

Method for verifying the functionality of a digital circuit

Embodiments herein describe techniques for synchronizing LFSRs located on two compute devices. To synchronize the LFSRs, a first one of the compute devices may transmit a first training block that includes a predefined bit sequence. The training block is scrambled by a transmitting (TX) LFSR on the first compute device and then transmitted to the second compute device. The second compute device performs an XOR operation to recover the outputs of the TX LFSR that were used to scramble the data. The second compute device can use the outputs of the TX LFSR to determine future outputs of the TX LFSR. These future outputs are then used to initialize a receiving (RX) LFSR on the second compute device. Now, when subsequent training blocks are received, the second compute device can use the initialized RX LFSR to descramble the scrambled training blocks.

Initializing a descrambler

In one or more embodiments, one or more systems, devices, methods, and/or processes described can send, via an interconnect, a rate master command to at least one of multiple processing nodes; determine that a message indicating a dropped command, associated with the rate master command, is received; determine that a count, associated with dropped commands, satisfies a threshold; and provide, to the processing nodes via the interconnect, a signal indicating a command rate, in response to determining that the count satisfies the threshold. Moreover, the count can be incremented in response to determining that the message is received. The at least one of multiple processing nodes can receive, via the interconnect, the signal indicating the command rate and can utilize the command rate in issuing speculative commands, via the interconnect.

Determining command rate based on dropped commands

Growing community of inventors

Rochester, MN, United States of America

Paul Allen Ganfield