Gate matrix layout problem plays an important role in integrated circuit design, but its optimization is NP-hard. In this paper, typical gate layout problem is analysed and adapted to neural network representation, fu...Gate matrix layout problem plays an important role in integrated circuit design, but its optimization is NP-hard. In this paper, typical gate layout problem is analysed and adapted to neural network representation, furthermore the simulated results are given.展开更多
双射映射(bijective mapping)能够实现二维基于旋转不变子空间的信号参数估计算法(two-dimensional estimating signal parameters via rotational invariance techniques,2D-ESPRIT)的参数自动配对,从而有效降低计算复杂度.在双射映射...双射映射(bijective mapping)能够实现二维基于旋转不变子空间的信号参数估计算法(two-dimensional estimating signal parameters via rotational invariance techniques,2D-ESPRIT)的参数自动配对,从而有效降低计算复杂度.在双射映射中,相较于其他部分,矩阵的实数化涉及规模更大、复杂度更高的计算,对系统实时性的影响更大.为优化这一过程,从矩阵的稀疏性出发,对其计算流程进行了深入分析,在FPGA平台上设计并实现了一种由通用与延长流水线构成的矩阵实数化硬件加速系统.实验结果表明:该系统的计算误差控制在1.14×10^(-4)以内;当总阵元数为6、快拍数为32时,该系统仅用时2.62μs完成计算,且达到最大加速比8.17;在不同快拍数和阵元数的情况下,系统的资源消耗保持稳定.展开更多
基金Support by Science Foundation of the Ministry of Posts and Telecommunications
文摘Gate matrix layout problem plays an important role in integrated circuit design, but its optimization is NP-hard. In this paper, typical gate layout problem is analysed and adapted to neural network representation, furthermore the simulated results are given.
文摘双射映射(bijective mapping)能够实现二维基于旋转不变子空间的信号参数估计算法(two-dimensional estimating signal parameters via rotational invariance techniques,2D-ESPRIT)的参数自动配对,从而有效降低计算复杂度.在双射映射中,相较于其他部分,矩阵的实数化涉及规模更大、复杂度更高的计算,对系统实时性的影响更大.为优化这一过程,从矩阵的稀疏性出发,对其计算流程进行了深入分析,在FPGA平台上设计并实现了一种由通用与延长流水线构成的矩阵实数化硬件加速系统.实验结果表明:该系统的计算误差控制在1.14×10^(-4)以内;当总阵元数为6、快拍数为32时,该系统仅用时2.62μs完成计算,且达到最大加速比8.17;在不同快拍数和阵元数的情况下,系统的资源消耗保持稳定.
