60results about How to "Implement parallel computing" patented technology

The invention relates to a method for carrying out parallelization numerical simulation on hydrodynamic force conditions of a river provided with a cascade hydropower station. According to the method, a closed equation set is established for each course by setting an upper level or a flow-level relation of a river channel structure as a lower boundary condition of an upper reach and supposing that a lower level of the structure is used as an upper boundary condition of a lower reach; and each equation set only comprises variables of the same reach, a coefficient matrix formed by adopting an implicit difference scheme to carry out dispersion are still kept having the banded characteristics and the physically related reaches are relatively independent on the aspect of value, so that the variables of the reaches can be solved by applying the banded matrix and the aim of ensuring the flows at both sides of the structure equal is fulfilled by iterating and correcting the lower level of the structure. Compared with the conventional method of solving the hydrodynamic conditions of the river by the single course, the method disclosed by the invention has the advantages that the rapid parallel solution of the multi-course river hydrodynamic conditions can be implemented without modifying an original sequence program of solving the river hydrodynamic conditions on a large scale; the operation process has short time consumption; and the efficiency is high.

The present invention provides a data processing method and equipment based on an AI chip. The method of the present invention divides the AI ​​chip data processing pipeline into the following three stages of processing: data acquisition and preprocessing, neural network model processing, and neural network model processing. Post-processing; the three-stage processing is a parallel pipeline structure; the AI ​​chip includes at least a first processor, a second processor and a third processor, the first processor is used for data acquisition and pre-processing, and the third processor is used for neural processing. Network model processing, the second processor is used for neural network model post-processing; the first processor, the second processor and the third processor perform the three-stage processing at the same time, which reduces the waiting time of the processors and maximizes the realization of It improves the parallel computing of each processor, improves the efficiency of AI chip data processing, and thus can increase the frame rate of AI chip.

The invention provides a brain-like computing chip and a computing device. The brain-like computing chip includes a many-core system composed of one or more functional cores, and data transmission is performed between the functional cores through an on-chip network; the functional cores include : at least one neuron processor for calculating multiple neuron models; at least one coprocessor, coupled with the neuron processor, for performing integral operations and / or multiply-add operations; wherein the neuron The metaprocessor may invoke the coprocessor to perform multiply-add-like operations. The brain-like computing chip provided by the invention is used to efficiently implement various neuromorphic algorithms, especially for the computing characteristics of the rich SNN model, synaptic computing with high computing density, and cell body computing with high flexibility requirements.

The invention discloses a parallel computation management-based autonomous navigation simulation and scheduling management system, which consists of a hardware system and a software system, wherein the hardware system consists of a master control computer, a real-time scheduling computer, a simulation computing node, a real-time communication network and a clock card; the real-time communication network consists of a giga Ethernet network and an infiniband high-performance real-time network; and the software system consists of a supporting operating system, master control end software and real-time scheduling control software. The parallel computation management-based autonomous navigation simulation and scheduling management system ensures the real-time property of a simulation system due to the adoption of the infiniband high-performance real-time network and an REDHAWK real-time operating system, ensures the possibility that the system realizes parallel computations in terms of hardware due to the adoption of a computing server and a blade server, realizes the parallel computations among simulation system models based on a data driving asynchronous pipeline parallel real-time scheduling algorithm, and in combination with the hardware, greatly speeds up the computation of the simulation system and ensures the parallelism and the real-time property of the complex simulation system.

The embodiment of the invention provides a method and a device for embedding a relation network graph in a multi-dimensional space through a distributed architecture. The distributed architecture comprises a plurality of processing devices, wherein one set of nodes of the relation network graph is distributed to each processing device. The method is executed by a certain processing device, and comprises the steps of: determining first-order adjacent nodes of a first node for which the processing device is responsible, determining processing devices corresponding to the first-order adjacent nodes; determining an initial embedding vector of the first node as a current order embedding vector; employing the current order embedding vector of the first-order adjacent nodes to calculate the nextorder embedding vector of the first node, and employing the next order embedding vector to update the current order embedding vector; and continuously executing the updating step until the current order quantity reaches a preset search depth, and taking the current order embedding vector at present as an embedding vector of the multi-dimensional space. Therefore, the distributed graph embedding method and device can effectively embed the relation network graph into the multi-dimensional space.