排序方式: 共有4条查询结果,搜索用时 15 毫秒
1
1.
2.
3.
随着AI芯片制程不断升级以及结构设计不断优化,AI芯片在并行计算能力和泛化能力等方面展现出明显的优势,这些优势使得AI芯片在宇航领域得到越来越多的关注和应用。与传统的质量保证方法不同,AI芯片测评作为一项新的重要环节,对于AI芯片的选型和使用具有重要意义。本文针对面向宇航应用的AI芯片,研究了一种宇航用AI芯片的测评方法,该方法可以测评出AI芯片的精度、算力、功耗和量化支持能力,并基于该方法对AI芯片的性能和功能性指标进行了详细的分析和评价,最后针对宇航用AI芯片测评方法的进一步研究方向提出了相关建议。 相似文献
4.
This paper combines improved Hamming codes and parity codes to assure the reliability of memory in presence of multiple bit upsets with low cost overhead.The redundancy bits of improved Hamming codes will be appended at the end of data bits,which eliminates the overhead of interspersing the redundancy bits at the encoder and decoder.The reliability of memory is further enhanced by the layout architecture of redundancy bits and data bits.The proposed scheme has been implemented in Verilog and synthesized using the Synopsys tools.The results reveal that the proposed method has about 19% less area penalties and 13% less power consumption comparing with the current two-dimensional error codes,and its latency of encoder and decoder is 63% less than that of Hamming codes. 相似文献
1