A data processing device includes: multiple data processing stages including a processing element, a stage memory and an event controller; and a bidirectional slotted bus connecting between the data processing stages, including two write only busses arranged at different data writing directions independently from each other. The processing element and the stage memory in one data processing stage are connected to each other via a read only bus. The processing element and the slotted bus are connected to each other via a write only bus. A process completion event is input from the processing element to the event controller, and an external event is input from an external device to the event controller. The event controller generates a task start event with respect to the processing element, according to each of the process completion event and the external event.

A data processing device includes: multiple data processing stages including a processing element, a stage memory and an event controller; and a bidirectional slotted bus connecting between the data processing stages, including two write only busses arranged at different data writing directions independently from each other. The processing element and the stage memory in one data processing stage are connected to each other via a read only bus. The processing element and the slotted bus are connected to each other via a write only bus. A process completion event is input from the processing element to the event controller, and an external event is input from an external device to the event controller. The event controller generates a task start event with respect to the processing element, according to each of the process completion event and the external event.

Method and apparatus for implementing an optimized credit return mechanism for packet sends. A Programmed Input/Output (PIO) send memory is partitioned into a plurality of send contexts, each comprising a memory buffer including a plurality of send blocks configured to store packet data. A storage scheme using FIFO semantics is implemented with each send block associated with a respective FIFO slot. In response to receiving packet data written to the send blocks and detecting the data in those send blocks has egressed from a send context, corresponding freed FIFO slots are detected, and a lowest slot for which credit return indicia has not be returned is determined. The highest slot in a sequence of freed slots from the lowest slot is then determined, and corresponding credit return indicia is returned. In one embodiment an absolute credit return count is implemented for each send context, with an associated absolute credit sent count tracked via software that writes to the PIO send memory, with the two absolute credit counts used for flow control.

Method and apparatus for implementing an optimized credit return mechanism for packet sends. A Programmed Input/Output (PIO) send memory is partitioned into a plurality of send contexts, each comprising a memory buffer including a plurality of send blocks configured to store packet data. A storage scheme using FIFO semantics is implemented with each send block associated with a respective FIFO slot. In response to receiving packet data written to the send blocks and detecting the data in those send blocks has egressed from a send context, corresponding freed FIFO slots are detected, and a lowest slot for which credit return indicia has not be returned is determined. The highest slot in a sequence of freed slots from the lowest slot is then determined, and corresponding credit return indicia is returned. In one embodiment an absolute credit return count is implemented for each send context, with an associated absolute credit sent count tracked via software that writes to the PIO send memory, with the two absolute credit counts used for flow control.

Embodiments of the present disclosure generally relate to a cloud computing network and a method of transferring information among processing nodes in a cloud computing network. In one embodiment, a cloud computing network is disclosed herein. The cloud computing network includes a plurality of motherboards arranged in racks. Each individual motherboard includes a central hub and a plurality of processing nodes equipped to the central hub. Each processing node is configured to access memory or storage space of another processing node in the same motherboard by intermediation of the hub. The access is called a communication between a pair of processing nodes. The communication includes a string of information transmitted between processing nodes. The string of information has a plurality of frames. Each frame includes a plurality of time slots, wherein each time slot is allotted a specific node pair.

Embodiments of the present disclosure generally relate to a cloud computing network and a method of transferring information among processing nodes in a cloud computing network. In one embodiment, a cloud computing network is disclosed herein. The cloud computing network includes a plurality of motherboards arranged in racks. Each individual motherboard includes a central hub and a plurality of processing nodes equipped to the central hub. Each processing node is configured to access memory or storage space of another processing node in the same motherboard by intermediation of the hub. The access is called a communication between a pair of processing nodes. The communication includes a string of information transmitted between processing nodes. The string of information has a plurality of frames. Each frame includes a plurality of time slots, wherein each time slot is allotted a specific node pair.

A group of transistors is configured to drive a bus at time slots, to express data on the bus. The group of transistors dissipates an amount of electrical energy when driving the bus to a logic level opposite to a logic level present on the bus in an immediate preceding time slot. The group of transistors is arranged to dump another amount of electrical energy. Dumping of the other amount of electrical energy is responsive to driving the bus to a logic level that is the same as present on the bus in an immediate preceding time slot. The dumped amount of electrical energy is equivalent to the amount of energy dissipated by the transistors when transitioning the bus to a different logic level. Other aspects are also described.

A group of transistors is configured to drive a bus at time slots, to express data on the bus. The group of transistors dissipates an amount of electrical energy when driving the bus to a logic level opposite to a logic level present on the bus in an immediate preceding time slot. The group of transistors is arranged to dump another amount of electrical energy. Dumping of the other amount of electrical energy is responsive to driving the bus to a logic level that is the same as present on the bus in an immediate preceding time slot. The dumped amount of electrical energy is equivalent to the amount of energy dissipated by the transistors when transitioning the bus to a different logic level. Other aspects are also described.

Reducing power consumption of communication interfaces by clock frequency scaling and adaptive interleaving of polling is disclosed. In a first aspect, a control system controls transmission of a command via a serial interface at a higher clock frequency. After transmission, the control system and the interface are operated at a lower clock frequency to save power during command execution. In this aspect, a reduction in polling corresponds to the reduction in clock signal frequency. When the command is complete, the interface is operated at the higher frequency to send another command. In a second aspect, after the control system sends a command to the receiving device, polling is suspended and an execution time of the command is tracked. Polling begins when the tracked execution time almost equals an expected completion time. Both aspects disclosed above may be implemented to reduce power consumption in exchange for a small increase in latency.

Methods and systems for shared control of a phase locked loop (PLL) for a multi-port physical layer (PHY) are disclosed. In one aspect, an arbitration logic circuit is coupled to ports of a multi-port PHY sharing a phase locked loop (PLL). Upon receiving an indication that the shared PLL is to be reset, the arbitration logic circuit commands the ports sharing the PLL to enter a state in which any reset of the shared PLL would have minimal or no effect in their operations. In this manner, an integrated circuit (IC) including a multi-port PHY may be configured with only one PLL and associated clock generating logic to provide a clock signal for some or all of its ports, thus reducing its semiconductor area and power consumption. Furthermore, the ports of the multi-port PHY may operate independently from each other obviating any configuration and/or interoperability problems associated with having a shared PLL.