3results about How to "Reduce synchronization overhead" patented technology

The invention relates to the technical field of communication network system simulation, and provides a distributed simulation method and system based on satellite-ground communication, a medium and electronic equipment. The distributed simulation method comprises the following steps: a pre-allocation step: allocating a plurality of satellites to be simulated to a plurality of logic processes, and maintaining a simulation model of at least one satellite in each logic process; in each logic process, pre-creating and maintaining a preset number of terminal simulation resources for each satellite; a simulation operation and switching step: when it is monitored that the terminal needs to be switched from a source logic process where the current satellite is located to a target logic process where a target satellite is located, the source logic process extracts simulation context information of the terminal and sends the simulation context information to the target logic process; and the target logic process selects idle terminal simulation resources from the terminal simulation resources, loads the simulation context information and activates the simulation context information. Through the setting, the problem of high overhead of cross-logic process communication is effectively solved, and the simulation efficiency is improved.

The invention discloses a large model distributed training method and system based on all-in-one machine hybrid cloud, and relates to the technical field of resource management. A hybrid cloud training architecture and secure communication are established between a local all-in-one machine cluster and a public cloud elastic cluster; dividing training data according to sensitivity, and realizing local storage of core sensitive data and cloud side storage of non-sensitive data; dividing the training task into a core task and an extension task, and scheduling the core task and the extension task to a local training node and a cloud side training node for execution; monitoring local resource load and task progress, and elastically applying and releasing cloud side training nodes according to training stages; and limiting cross-domain transmission as necessary information, updating model parameters through gradient aggregation, and synchronously entering the next iteration. Through the technical scheme of the invention, on the premise of ensuring that sensitive data is not out of a domain, on-demand expansion and recovery of cloud side computing power are realized, cross-domain communication overhead is reduced, distributed training throughput and stability are improved, a training period is shortened, and comprehensive cost is reduced.

The application discloses a heterogeneous parallel material removal calculation method and system. The method comprises the following steps: constructing a heterogeneous parallel calculation environment comprising at least a first processing unit and a second processing unit, wherein at least one processing unit supports data parallel execution for a large number of discrete elements; constructing a discrete representation of a workpiece, and generating state data for representing a material state based on the discrete representation of the workpiece; generating space constraint information based on tool geometric parameters, relative motion information of the tool and the workpiece, and the state data; decomposing material removal simulation calculation into a preprocessing task and a relationship evaluation task; mapping at least two calculation tasks to the first processing unit and the second processing unit for execution, so that the first processing unit and the second processing unit concurrently or flow water overlap execute in at least part of a time period; updating the material state based on the space relationship quantity and calculating the material removal quantity, and iterating until the trajectory is completed. Therefore, the calculation efficiency is significantly improved, and the real-time simulation requirement is met.