Methods and apparatus for implementing adaptive equalization channel extension retimer link-up in high-speed serial links. Under aspects of the proposed methodology, the retimer device intervenes with the adaptive equalization training, training both ends of the link individually between the two end points, and once both of the retimer's receivers are trained, it propagates the receiver readiness indication through from one end point to the other. This allows all sections of the link to train at the same time, and for all devices to transition to data mode at the same time, once all channels have been adapted to and trained. This methodology also applies to cascaded retimer device configurations, where multiple retimers are being used to extend the channel even further.

A signal routing circuit is disclosed which employs resistive combiners to reduce signal jitter. A signal routing circuit configured according to an embodiment comprises an input stage including a plurality of buffer circuits. Each of the buffer circuits is controlled by a selection signal to enable an input signal at an input port of the buffer circuit to generate an output signal at an output port of the buffer circuit. The signal routing circuit also includes a plurality of resistors to couple the output port of each of the buffer circuits of the input stage to a summing junction. The signal routing circuit further includes an output stage including an additional buffer circuit. The input port of the additional buffer circuit is coupled to the summing junction, and the output port of the additional buffer circuit is configured to provide the routed output signal based on the selection signals.

Disclosed example methods for scheduling a computer-executable job include determining valid combinations of nodes and links from a network of nodes interconnected by links, the valid combinations of nodes and links having capability and capacity to complete the computer-executable job; determining, from the valid combinations of nodes and links, first combinations of nodes and links based on a total cost associated with execution of the computer-executable job on corresponding ones of the valid combinations of nodes and links; selecting one of the first combinations of nodes and links based on a load balancing status between the first combinations of nodes and links; and scheduling the computer-executable job to be executed by the one of the first combinations of nodes and links.

A method is provided for operating a communication controller coupling a device comprising a processor with a bus. The method comprises: receiving a plurality of types of data packets via the bus and processing received data packets before making available said received data packets to the device processor. The processing of received data packets comprises: evaluating each received data packet in accordance with predetermined criteria; rejecting any of the received data packets that fails to meet the predetermined criteria; identifying non-rejected data packets having high priority; identifying said non-rejected other data packets having lower priority; providing a high priority data path to the processor for the high priority data packets; providing at least one additional data path to the processor for the other data packets; and providing a high priority alert to the device processor to the presence of high priority data packets at the high priority channel.

Disclosed example methods for scheduling a computer-executable job include determining valid combinations of nodes and links from a network of nodes interconnected by links, the valid combinations of nodes and links having capability and capacity to complete the computer-executable job; determining, from the valid combinations of nodes and links, first combinations of nodes and links based on a total cost associated with execution of the computer-executable job on corresponding ones of the valid combinations of nodes and links; selecting one of the first combinations of nodes and links based on a load balancing status between the first combinations of nodes and links; and scheduling the computer-executable job to be executed by the one of the first combinations of nodes and links.

In a communication apparatus, a buffer offset calculator calculates a buffer offset for a transmission-ready state k-th communication unit in accordance with a total number of at least one of data packets that is (i) transmittable by at least one m-th communication unit in the N communication units except for the k-th communication unit and (ii) completely received by the recipient device within a data transmission period of the communication network corresponding to the k-th communication unit (m is each of variables 1 to N except for k). A data packet allocator selects, in a queue of the data packets stored in the transmission buffer, one of the data packets, a beginning of the selected data packet being located offset by the buffer offset relative to a beginning of the queue of the data packets. The data packet allocator allocates the selected data packet to the k-th communication unit.

In a communication apparatus, a buffer offset calculator calculates a buffer offset for a transmission-ready state k-th communication unit in accordance with a total number of at least one of data packets that is (i) transmittable by at least one m-th communication unit in the N communication units except for the k-th communication unit and (ii) completely received by the recipient device within a data transmission period of the communication network corresponding to the k-th communication unit (m is each of variables 1 to N except for k). A data packet allocator selects, in a queue of the data packets stored in the transmission buffer, one of the data packets, a beginning of the selected data packet being located offset by the buffer offset relative to a beginning of the queue of the data packets. The data packet allocator allocates the selected data packet to the k-th communication unit.

An improved method of hierarchical output queueing of packets for a network scheduler of a network gateway that determines delays needed to conform to applicable rate shaping, and enqueues the packets based on the delay. Queues are associated with different classes of service (CoS), and within each class, each queue has a different scheduled time upon which it becomes available for dequeuing. A single set of CoS queues can support a large number of devices, improving the efficiency of software-based queuing by reducing the number of queues and simplifying queue polling.

According to an embodiment, an information processing apparatus includes a prefetch unit and a scheduler unit. The prefetch unit is configured to prefetch a scheduling entry corresponding a future time period in advance from scheduling information including one or more entries each of which at least contains a transmission state and interval for each of one or more transmission queues. The scheduler unit configured to determine a starting time of transmission for each frame waiting for transmission in each queue, on the basis of the prefetched entry.

A real-time Ethernet (RTE) protocol includes start-up frames originated by a master device for network initialization including a preamble, destination address (DA), source address (SA), a type field, and a status field including state information that indicates a current protocol state that the Ethernet network is in for the slave devices to translate for dynamically switching to one of a plurality of provided frame forwarding modes. The start-up frames include device Discovery frames at power up, Parameterization frames that distribute network parameters, and Time Synchronization frames including the master's time and unique assigned communication time slots for each slave device. After the initialization at least one data exchange frame is transmitted exclusive of SA and DA including a preamble that comprises a header that differentiates between master and slave, a type field, a status field excluding the current protocol state, and a data payload.