Hierarchical redundancy design for WSI neuro-processors
Nobuhiro Tomabechi
Abstract
Nobuhiro Tomabechi
Abstract
This paper proposes the hierarchical redundancy design for the defect recovery of wafer scale neuro-processors. The design combines both the circuit level redundancy and the system level redundancy. The hierarchical redundancy design results in that a neuro-processor composed of 500 neurons may be implemented on a single wafer with 27% increase of hardware and with sufficient yield.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
This paper proposes the hierarchical redundancy design for the defect recovery of wafer scale neuro-processors. The design combines both the circuit level redundancy and the system level redundancy. The hierarchical redundancy design results in that a neuro-processor composed of 500 neurons may be implemented on a single wafer with 27% increase of hardware and with sufficient yield.
Key concepts: Redundancy (engineering), Wafer-scale integration, Computer science, Wafer, Computer architecture, Embedded system, Parallel computing, Computer hardware