首页> 外国专利> PARALLEL COMPUTATION ARCHITECTURE WITH RECONFIGURABLE CORE LEVEL AND VECTOR LEVEL PARALLELITY

PARALLEL COMPUTATION ARCHITECTURE WITH RECONFIGURABLE CORE LEVEL AND VECTOR LEVEL PARALLELITY

机译:具有可重新配置核心级别和向量级并行度的并行计算架构

摘要

Neural network processing hardware using parallel computation architectures with reconfigurable kernel-level and vector-level parallelism is provided. In various embodiments, a working memory of a neural network model is adapted to store a neural network model that has a plurality of layers. Each layer has at least one dimension and has a plurality of synaptic weights. A plurality of neural cores is provided. Each neural core contains a computation unit and an activation work memory. The calculation unit is adapted to apply a plurality of synaptic weights to a plurality of input activations in order to generate a plurality of output activations. The calculation unit has a plurality of vector units. The activation memory is adapted to store the input activations and the output activations. The system is adapted to partition the plurality of cores into a plurality of partitions based on dimensions of the layer and the vector units.
机译:提供了使用具有可重新配置内核级别和矢量级并行性的并行计算架构的神经网络处理硬件。在各种实施例中,神经网络模型的工作存储器适于存储具有多个层的神经网络模型。每个层具有至少一个尺寸并且具有多个突触权重。提供多个神经核。每个神经核心包含计算单元和激活工作存储器。计算单元适于将多个突触权重应用于多个输入激活,以便生成多个输出激活。计算单元具​​有多个向量单元。激活存储器适于存储输入激活和输出激活。该系统适于基于层和向量单元的尺寸将多个核心分配到多个分区中。

著录项

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号