Design and Evaluation of a Carbon Nanotube-Based Programmable Architecture

In the hunt to find a replacement to CMOS, material scientists are developing a wide range of nanomaterials and nanomaterial-based devices that offer significant performance improvements. One example is the Carbon Nanotube Field Effect Transistor, or CNFET, which replaces the traditional silicon channel with an array of semiconducting carbon nanotubes (CNTs). Given the increased variation and defects of nanometer-scale fabrication, and the regular nature of bottom-up self-assembly, field programmable devices are a promising initial application for such technologies. In this paper, we detail the design and evaluation of a novel nanomaterial-based architecture called FPCNA (Field Programmable Carbon Nanotube Array). New nanomaterial-based circuit building blocks are developed and characterized, including a lookup table created entirely from continuous CNT ribbons. To accurately determine the performance of these building blocks, we create variation-aware physical design tools with statistical timing analysis that can handle both Gaussian and non-Gaussian random variables. When the FPCNA architecture is evaluated using this CAD flow, we see a 2.75× performance improvement over an equivalent CMOS FPGA at a 95% yield. In addition, FPCNA offers a 5.07× footprint reduction compared to the baseline FPGA.     

东北地区冬季气温对海温异常响应的研究

利用HadISST OI海温和中国东北地区92站逐日气温资料,使用广义平衡反馈分析方法(GEFA)结合EOF分析方法(GEFA-EOF)研究了近50 a中国东北地区冬季气温对海表温度异常(SSTA)的响应.结果表明:对于热带和北半球中纬度5个海盆来说,东北地区冬季气温异常与同期热带大西洋和北大西洋海温异常有密切关系,与其他海盆关系不显著;热带大西洋的"正—负—正"三极型模态(TA3)以及北大西洋纬向上"正—负—正"三极型模态(NA3)分别对东北地区冬季气温的异常偏低和偏高有显著的强迫作用,且对北部地区的强迫作用大于南部地区.热带大西洋和北大西洋对东北地区冬季气温异常影响的可能途径为:热带大西洋TA3模态通过在北半球激发的"正—负—正"的遥相关波列,致使东亚大槽移至贝加尔湖地区,有利于极地冷空气南下至东北地区,导致该地区的冷冬;北大西洋的"正—负—正"三极型模态(NA3)直接响应使得东亚大槽减弱消失,极地冷空气南下受阻,导致该地区冬季气温异常偏高.     

The Role of Warm North Atlantic SST in the Formation of Positive Height Anomalies over the Ural Mountains during January 2008

The most severe snowstorm and freezing-rain event in the past 50 years hit central and southern China in January 2008. One of the main reasons for the anomalous climate event was the occurrence of atmospheric circulation anomalies over middle and high latitudes, particularly the persistent blocking that occurred over the Ural Mountains. Along with atmospheric anomalies, a strong La Nina event in the Pacific and warm sea surface temperature anomalies (SSTAs) in the North Atlantic were the most significant in the lower boundary. Since a brief analysis suggests that La Nina exerts no significant impact on the Urals, the key point of focus in this study is on the role of the warmer SSTAs in the North Atlantic. Based on an observational composite, North Atlantic SSTAs pattern when the height anomaly over the Urals is strongly positive is found similar to that in January 2008, but no significant SSTAs occurred elsewhere, such as the Pacific. Using an atmospheric general circulation model, ECHAM5, the impact of North Atlantic SSTAs on the extratropical atmosphere circulation in the event was investigated. The results show that the warm SSTAs strengthened the blocking high over the Urals, through anomalous transient eddies. The consistency between the study model and the observational composite indicates that the warm SSTAs in the North Atlantic were indeed an important factor in the formation of the snowstorm disaster of January 2008.     

Statistical static timing analysis for circuit aging prediction

Complementary metal oxide semiconductor ( CMOS) aging mechanisms including bias temperature instability ( BTI) pose growing concerns about circuit reliability. BTI results in threshold voltage increases on CMOS transistors, causing delay shifts and timing violations on logic circuits. The amount of degradation is dependent on the circuit workload, which increases the challenge for accurate BTI aging prediction at the design time. In this paper, a BTI prediction method for logic circuits based on statistical static timing analysis (SSTA) is proposed, especially considering the correlation between circuit workload and BTI degradation. It consists of a training phase, to discover the relationship between circuit scale and the required workload samples, and a prediction phase, to present the degradations under different workloads in Gaussian probability distributions. This method can predict the distribution of degradations with negligible errors, and identify 50% more BTI-critical paths in an affordable time, compared with conventional methods.     

一种基于区域海表面温度异常预测的ENSO预报深度学习模型

本文通过实验,从设计的四种深度学习网络模型中挑选出一种可以用来预测区域海平面温度异常 (SSTA) 的模型——序列到序列 (Seq2Seq) 模型,并确定了适合于此模型的一组最优的超参数 (编码器使用正序输入、不使用 L2 正则化和使用 Adam 优化器) 和输入变量个数。与传统的动力学 ENSO 预报模型相比,该模型在中长期 (提前 7 个月以上) 预测上的均方根误差 (RMSE) 表现要更好。在实际的 SSTA (Niño3.4 指数) 预测实验中,该模型可以较好地预测出 SSTA 变化趋势,但在峰值处表现较差。与其他动力学和统计模型相比,该模型有较好的预测结果。因此,考虑其在中长期优于动力学模型的表现以及整体较好的预测结果,该模型可以作为一种 ENSO 预报以及区域 SSTA 预测的深度学习模型。     

一种考虑空间关联工艺偏差的统计静态时序分析方法

为了准确评估工艺参数偏差对电路延时的影响,该文提出一种考虑空间关联工艺偏差的统计静态时序分析方法。该方法采用一种考虑非高斯分布工艺参数的二阶延时模型,通过引入临时变量,将2维非线性模型降阶为1维线性模型;再通过计算到达时间的紧密度概率、均值、二阶矩、方差及敏感度系数,完成了非线性非高斯延时表达式的求和、求极大值操作。经ISCAS89电路集测试表明,与蒙特卡洛仿真(MC)相比,该方法对应延时分布的均值、标准差、5%延时点及95%延时点的平均相对误差分别为0.81%, -0.72%, 2.23%及-0.05%,而运行时间仅为蒙特卡洛仿真的0.21%,证明该方法具有较高的准确度和较快的运行速度。     

利用GPU加速基于稀疏网格的SSTA

提出一种利用图形处理单元（Graphics Processing Unit,GPU）加速统计静态时序分析的方法,利用稀疏网格减少统计静态时序分析中时序图各节点的配置个数,在GPU上构建复杂的时序图数据结构后并行计算各节点的不同配置,达到加速统计静态时序分析的目的。测试结果表明,提出的方法能够在不损失精度的前提下,将统计静态时序分析运行速度平均提高300倍以上。随着现代集成电路规模的持续增大和集成电路工艺的不断发展,这种新型快速的统计静态时序方法能够有效提高时序分析的速度和效率。     

基于自适应差分进化算法优化极限学习机的干旱预测方法

针对极限学习机在实际应用时随机选取初始权值与阈值易导致其稳定性弱及泛化能力差的问题,利用自适应差分进化算法对其进行改进,构建了自适应差分进化极限学习机预测模型,并选用海表异常温度作为该模型的输入因子,对研究区域的干旱进行预测。结果表明,以海表异常温度作为模型的输入因子,应用极限学习机能有效地进行干旱预测,通过自适应差分进化算法优化的极限学习机应用于干旱预测,其精度与稳定性均有所提高。