Interconnect reliability analysis of ULSI using automated model generation algorithm

Lin, Qian; Fu, Haipeng; Na, Weicong; He, Feifei; Li, Xi; Cheng, Qian‐Fu; Zhu, Yi

doi:10.1002/mmce.20992

Cited by 9 publications

(3 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Step 3) Each processor reads the corresponding file, i.e., the ith processor reads the file containing N l,i training data, then computes E k train,i and ∂E k train,i ∂w using (16) and (17).…”

Section: Parallel Ann Trainingmentioning

confidence: 99%

“…The computationally intensive and time-consuming ANN training during adaptive sampling process in [14] is replaced by simple and efficient interpolation computation, thus making AMG faster than that in [14]. This modified AMG algorithm [15] has been applied in microwave power amplifier modeling [16] and interconnect reliability analysis [17]. However, the existing AMG with interpolation approaches in [15] is based on a sequential computation mechanism, involving sequential interpolation calculation for all subregions during adaptive data sampling, sequential data generation by repetitively driving detailed EM/physics/circuit simulators, and batch ANN model training through iterative training stages.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Automated Neural-Based Modeling of Microwave Devices Using Parallel Computation and Interpolation Approaches

Zhang

Yan

et al. 2019

IEEE Access

Self Cite

View full text Add to dashboard Cite

Automated model generation (AMG) is an automated artificial neural network (ANN) modeling algorithm, which integrates all the subtasks (including adaptive sampling/data generation, model structure adaptation, training, and testing) in neural model development into one unified framework. In existing AMG, most of the time is spent on data sampling and model structure adaptation due to the iterative neural network training and the sequential computation mechanism. In this paper, we propose an advanced AMG algorithm using parallel computation and interpolation approaches to speed up the neural modeling of microwave devices. Efficient interpolation approaches are incorporated to avoid repetitive training of the intermediate neural networks during adaptive sampling process in AMG. Parallel computation formulation based on a multi-processor environment is proposed to further save time during interpolation calculation, data generation, and model structure adaptation process. Examples of automated modeling of two microwave filters are presented to show the advantage of this paper. INDEX TERMS Design automation, modeling, neural networks, parallel computation, interpolation approaches.

show abstract

Section: Parallel Ann Trainingmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Automated Neural-Based Modeling of Microwave Devices Using Parallel Computation and Interpolation Approaches

Zhang

Yan

et al. 2019

IEEE Access

Self Cite

View full text Add to dashboard Cite

show abstract

“…It is reported that failure rate caused by EM is extremely high after 0.18um technology node in Al [17]. Nowadays, EM can be predicted by combing the FEA and ANN techniques for ICs [18], [19]. In addition, the introduction of new materials and processing schemes leads to even more challenges in guaranteeing interconnect robustness.…”

Section: Volume 8 Number 3 July 2019mentioning

confidence: 99%

Study of the Interconnect Failure Mechanism and Micro-effort for ULSI

Lin¹,

Wu²

2019

IJCCE

Self Cite

View full text Add to dashboard Cite

Interconnect reliability has been regarded as a discipline that must be seriously taken into account from the early design phase of ultra large scale integration (ULSI). A synthetic review of valuable solutions to improve interconnect reliability is proposed in this paper. At first, a comprehensive review of the interconnect failure mechanisms and micro-efforts are carried out. Four types of interconnect failure mechanisms including EM, SM, TM and TDDB are illustrated in detail. Depending on their different effects for failure, the interconnect micro-effects are classified into the positive effects including the short-length effect, self-healing effect, reservoir effect and the negative effects which involves the skin effect, joule heating, current crowding , bandwidth, coupled noise, parasitic, RC delay, crosstalk and power dissipation and so on. Secondly, a novel qualitative evaluation method based on the radar chart has been presented, which visually shows the contrast of the key performance related to the interconnect failures. These present results might provide some valuable guidance for the study of IC interconnect reliability.

show abstract