7 results about "Bus routing" patented technology

The invention relates to the technical field of artificial intelligence, and discloses a hybrid bonding-based large model reasoning data read-write method and system, a controller and a storage medium, and the method comprises the following steps: constructing hybrid bonding connection between a PIM chip and a DRAM chip and between the PIM chip and a FLASH chip; when the large model is used for reasoning, the weight of the large model is stored in the FLASH chip through the data bus, and the cue word of the large model is stored in the DRAM chip through the data bus; in the pre-filling stage, the PIM chip is used for respectively reading the weight and the cue word, and the calculated first key value is cached and written into the DRAM chip; and in a decoding stage, respectively reading the weight and the first key value cache by utilizing the PIM chip, and writing the second key value cache of the Token generated by the large model into the DRAM chip. According to the method, the processing engine in the PIM chip is utilized, data only need to be read from the FLASH and the DRAM which are closest in physical distance, delay is reduced, model weight data and key value cache data do not need to be read through a bus route, data carrying is reduced, and power consumption is reduced.

The invention discloses a real-time task arbitration method of a heterogeneous OS cluster and related equipment, and relates to the technical field of computers, the method comprises the following steps: collecting a resource request signal, task priority data, a global timestamp and a resource identifier of a multi-core processor through a hardware interface; performing standardized mapping on task priority data, and dynamically calculating an effective priority based on waiting time; aiming at concurrent resource request signals of resources of the same resource identifier, arranging the resource request signals in a descending order according to effective priorities through a hardware sorting network, and performing timestamp defeat on the highest priority group by adopting a global timestamp; sending a resource authorization signal to the winning core and configuring a bus route; and dynamically adjusting a priority weighting parameter according to the historical arbitration data. According to the method, various data are collected through a hardware interface, then calculation is effective, then the global timestamp is adopted to carry out decision arbitration, and a priority hardware-assisted cross-OS real-time arbitration scheme can be realized.

The invention relates to the technical field of flight simulation communication, and discloses a flight simulation hierarchical communication system and method based on cloud native dual-bus scheduling, and a communication architecture based on hierarchical decoupling, and the system comprises a component application module, a bus intelligent scheduling module, a data distribution module and a transmission adaptation module which are connected in sequence. The component application module is used for docking various flight simulation components; the bus intelligent scheduling module comprises a control bus unit and a data bus unit which are physically isolated from each other, and an intelligent scheduling unit which is in communication connection with the control bus unit and the data bus unit, and is used for realizing type identification of transmission data, branch and bus routing and dynamic scheduling of transmission resources based on a flight simulation scene; and the data distribution module comprises a cloud resource adaptation engine and a subscription relationship management unit, and is used for docking a cloud platform resource pool and realizing heterogeneous deployment resource matching and communication channel subscription management of various flight simulation components.

The application discloses a bus wiring method based on GoldenBit and Hanan grid updating, which comprises the following steps: firstly, generating the GoldenBit topology path of each component, and then wiring other pin feet; then, before each wiring, performing dynamic Hanan grid updating; for each edge of the Hanan grid, according to whether it is a segment and a line extended from a component, the edge is divided into different priorities; finally, according to the different priorities, adjusting the wiring path, finding the shortest path while ensuring that the topology path is as same as the last path as possible, so as to complete the bus wiring.

This application provides a bus routing method for integrated circuits, an electronic device, and a computer-readable storage medium, relating to the field of integrated circuits, to solve the problem of time-consuming and labor-intensive bus routing in integrated circuits. The method includes: displaying a first interface, the first interface including a first control for indicating project creation. In response to a first operation on the first control, a project is created, and a second interface is displayed; the created project includes pins to be connected in the integrated circuit and the connection relationships of the pins to be connected; the second interface includes information on multiple groups and a second control; the multiple groups are obtained by automatically pre-grouping multiple connections, and the multiple connections are obtained according to the connection relationships of the pins to be connected. In response to a second operation on the second control, the result of automatic routing of the connections included in each group is output. This method is used to improve the efficiency of bus routing.

The invention discloses an event bus dynamic routing distribution method, device and equipment and a storage medium, relates to the field of computers, is applied to a terminal carrying a dynamic routing distribution system, and comprises the following steps: obtaining a target event, and determining a multi-dimensional feature corresponding to the target event by using a target feature extraction model; based on the multi-dimensional features of the target event and the real-time state of the dynamic routing distribution system, determining a routing path corresponding to the target event by using a target decision model, and obtaining a theoretical quantification result corresponding to each routing path by using a target reward function; the routing path is a known routing path corresponding to the event type of the target event; and determining a target routing path from all routing paths corresponding to the target event based on the theoretical quantification result, and distributing the target event to a target event queue corresponding to the target routing path based on the target routing path. According to the method and the device, the event bus route distribution mode is improved, and the route distribution requirement in a high-concurrency and complex dynamic scene is met.

The invention discloses an anti-collision control circuit and method for a wireless communication signal, and relates to the technical field of wireless communication, the anti-collision control circuit comprises an RISC-V core processor, a bus routing circuit, an APB bus, a digital baseband circuit, an RIB bus, a read-only memory and a random access memory, the bus routing circuit is respectively connected with the RISC-V core processor, the APB bus and the RIB bus, the RIB bus is respectively connected with the read-only memory and the random access memory, and the APB bus is connected with the digital baseband circuit; the RISC-V core processor comprises an instruction fetching circuit, a decoding circuit and an execution circuit; the digital baseband circuit comprises a sending link circuit, a receiving link circuit, a multiplexing circuit, a baseband controller and an anti-collision algorithm hardware acceleration circuit. According to the method, the calculation delay is remarkably reduced while the algorithm is ensured to be flexible and configurable, and the identification efficiency and the system throughput performance in a multi-label dense scene are improved.