International Business Machines Corporation

Qualcomm Incorporated

Robert Michael Bunce

Brian Joel Schuh

Eric M Schwarz

Hung Cai Ngo

Leon Jacob Sigal

Mark R Bilak

Steven C Parker

Christos J Georgiou

Valentina Salapura

Carrie Ellen Cox

Kevin G Kramer

Michael Raymond Trombley

Timothy Michael Kemp

Monty Montague Denneau

William Richard Kelly

Louis T Fasano

Thomas J Bardsley

Victor Git-Han Moy

Dinesh B Nair

Clock data recovery broadcast for multi-lane SerDes is disclosed. In some implementations, a serial input/output (I/O) interface includes a master lane and a plurality of slave lanes to receive serial incoming data. The master lane has a master clock and data recovery (CDR) module to generate master data sample clock control signals. The master data sample clock control signals can be broadcasted to the slave lanes. Furthermore, each of the plurality of slave lanes having a slave CDR module. The slave CDR module can include a clock edge tracking module to generate local data sample clock control signals, and a multiplexer to select the master data sample clock control signals or local data sample clock control signals to apply in the respective slave lane in response to a signal from a CDR controller indicative of a low power mode.

Clock data recovery broadcast for multi-lane SerDes

Systems and methods to provide a low-power mode for serial links are disclosed. One embodiment of such a system includes a transmitter coupled to a link; a receiver coupled to the link and configured to receive signals over the link from the transmitter; a transmit control module configured to cause the transmitter to enter and exit a low-power mode; and a clock module coupled to the transmitter and configured to provide a clock signal to the transmitter, wherein the clock module is further configured to provide the clock signal as a divided clock signal to the transmitter when the transmitter is in the low-power mode, further wherein the divided clock signal has a same phase as the clock signal before entry into the low-power mode.

Systems and methods providing a low-power mode for serial links

An apparatus for setting the timing of a triggering edge of a clock signal with respect to received parallel data. The apparatus includes a set of flip-flops including respective data inputs, respective clock inputs, and respective data outputs, wherein the set of flip-flops are configured to generate a set of output data at the data output based on parallel data applied to the respective data inputs in response to a triggering edge of a clock signal applied to the clock inputs; a variable delay element configured to apply a calibrated delay to the clock signal; and a controller configured to generate a control signal for the variable delay element to apply the calibrated delay to the clock signal based on the set of output data generated at the data outputs of the set of flip-flops.

Apparatus and method for centering clock signal in cumulative data eye of parallel data in clock forwarded links

An apparatus and method for dynamically allocating memory between inbound and outbound paths of a networking protocol handler so as to optimize the ratio of a given amount of memory between the inbound and outbound buffers is presented. Dedicated but sharable buffer memory is provided for both the inbound and outbound processors of a computer network. Buffer memory is managed so as to dynamically alter what portion of memory is used to receive and store incoming data packets or to transmit outgoing data packets. Use of the present invention reduces throttling of data rate transmissions and other memory access bottlenecks associated with conventional fixed-memory network systems.

Dynamic memory allocation between inbound and outbound buffers in a protocol handler

A method for on-chip detection of data lock and measurement of data lock time in a high-speed serial data link, including: permitting one or more incoming data streams into the high-speed data link; establishing a pattern to be searched in the one or more incoming data streams; comparing patterns in the one or more incoming data streams to a programmable data pattern; holding a repetitive pattern of bits in the one or more incoming data streams by one or more programmable data pattern registers, wherein when one or more occurrences of a byte are detected, an appropriate bit in the one or more programmable data pattern registers is set to indicate the byte's relative position; and filtering false indications in the repetitive pattern by using a byte detection state machine, the state machine controlling and keeping track of a search progress.

On-chip detection and measurement of data lock in a high-speed serial data link

An architecture that achieves high speed performance in a network protocol handler combines parallelism and pipelining in multiple programmable processors, along with specialized front-end logic at the network interface that handles time critical protocol operations. The multiple processors are interconnected via high-speed interconnect, and each processor's memory is globally accessible by other processors. Each processor has multiple threads, each capable of fully executing programs. Each processors contains embedded dynamic random access memory (DRAM). Threads within a processor are assigned the processing of various protocol functions in a parallel/pipelined fashion. Data frame processing is done by one or more of the threads to identify relates frames. Related frames are dispatch to the same thread so as to minimize the overhead associated with memory accesses and general protocol processing. The high-speed protocol handler may also provide built-in monitors for examining the activity of its hardware resources and reallocating the workload to the resources that are not heavily used, thus balancing the resource utilization and increasing the workload throughput.

Programmable network protocol handler architecture

A data communication controller processes incoming data frames. The controller includes a pre-processing block for receiving data frames and a frame processing unit coupled to the pre-processing block. The pre-processing block is configured to compare header fields of a current frame with header fields of a previous frame. The pre-processing block provides an output signal to the frame processing unit on the basis of the comparison of the header fields of the current and previous frames. The controller may operate in accordance with the Fiber Channel protocol, and the output signal may include bits to indicate that the current frame is of the same exchange, of the same sequence, and is next in sequence relative to the previous frame.

Inbound data stream controller with pre-recognition of frame sequence

A method and system for increasing the efficiency of packet processing within a packet protocol handler. In accordance with the method of the present invention packet processing tasks are performed on multiple processors or threads concurrently and in a pipelined fashion. Subsequent protocol packet processing tasks for processing a single packet are performed on multiple processors or threads, acting as stages of a pipeline. The assignment of tasks to processors or threads is performed dynamically, by checking the availability of a processor or thread in the subsequent pipeline stage. The availability determination includes determining the available capacity of the input work queue associated with each processor or thread. If the subsequent pipeline stage is overloaded, the task is assigned to another processor or thread that is not overloaded.

Pipelined packet processing

An integrated circuit real-time data tracing apparatus for analyzing microprocessor based computer systems for monitoring, in real-time, parameters sufficient to define the load and store operations information that the embedded core controller may assert, and process information during events. Integral on this single chip apparatus is a data trace unit designed to access control, address, and data signal lines required to monitor the embedded core controller's activities; perform data tracing independent of instruction tracing; synchronize with an instruction trace stream; allow for selection of multiple ranges for data tracing; report lost events to a FIFO array; and, output strobe signals to give a cycle accurate indication of when an event has been captured.

Integrated real-time data tracing with low pin count output

A system implementing a methodology for determining the exponent in parallel with determining the fractional shift during normalization according to partitioning the exponent into partial exponent groups according to the fractional shift data flow, determining all possible partial exponent values for each partial exponent group according to the fractional data flow, and providing the exponent by selectively combining possible partial exponents from each partial exponent group according to the fractional data flow. There is also provided a system implementing a methodology for generating the sticky bit during normalization. Sticky bit information is precalculated and multiplexed according to the fractional dataflow. In an embodiment of the invention, group sticky signals are calculated in tree form, each group sticky having a number of possible sticky bits corresponding to the shift increment amount of the multiplexing. The group sticky bits are further multiplexed according to subsequent shift amounts in the fractional dataflow to provide an output sticky bit at substantially the same time as when the final fractional shift amount is available, and thereby at substantially the same time as the normalized fraction.

Parallel calculation of exponent and sticky bit during normalization

A system implementing a methodology for determining the exponent in parallel with determining the fractional shift during normalization according to partitioning the exponent into partial exponent groups according to the fractional shift data flow, determining all possible partial exponent values for each partial exponent group according to the fractional data flow, and providing the exponent by selectively combining possible partial exponents from each partial exponent group according to the fractional data flow. There is also provided a system implementing a methodology for generating the sticky bit during normalization. Sticky bit information is precalculated and multiplexed according to the fractional dataflow. In an embodiment of the invention, group sticky signals are calculated in tree form, each group sticky having a number of possible sticky bits corresponding to the shift increment amount of the multiplexing. The group sticky bits are further multiplexed according to subsequent shift amounts in the fractional damflow to provide an output sticky bit at substantially the same time as when the final fractional shift amount is available, and thereby at substantially the same time as the normalized fraction.

Growing community of inventors

Hopewell Junction, NY, United States of America

Robert Michael Bunce