Abstract:This paper proposes a dynamic capacitance matching (DCM)-based RC current response algorithm for calculating the current waveform of a signal line without performing SPICE simulation. Specifically, unlike previous method such as CCS model, driver linear representation, waveform functional fitting or equivalent load capacitance, our algorithm does not rely on fixed reduced model of both standard cell driver and RC load. Instead, our algorithm approaches the current waveform dynamically by computing current responses of the target driver for various load scenarios. Besides, we creatively use symbolic expression to combine the y-parameter of RC network with the pre-characterized driver library in order to perform capacitance matching by considering over/under-shoot effect. Our algorithm is experimentally verified on 40nm CMOS technology and has been partially adopted by latest commercial tool for other nodes. Experimental results show that our algorithm has excellent resolution and promising efficiency compared with traditional methods and SPICE golden result, especially for application in computing delay, power and signal line electromigration.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to calculate the current waveform of the signal - line RC network without performing SPICE simulation. Specifically, existing methods such as CCS model, driver linear representation, waveform function fitting or equivalent load capacitance all rely on fixed simplified models of standard - cell drivers and RC loads. These methods have difficulty maintaining accurate current responses when dealing with complex RC loads, especially when considering overshoot/undershoot effects. Therefore, the paper proposes an RC current - response algorithm based on dynamic capacitance matching (DCM), aiming to dynamically approximate the current waveform by calculating the current response of the target driver in different load scenarios, and use symbolic expressions to combine the y - parameters of the RC network with the pre - characterized driver library to achieve capacitance matching.
### Main contributions of the paper:
1. **Dynamic Capacitance Matching (DCM)**: The paper proposes a new method that matches the actual response of the RC network by dynamically adjusting the capacitance value, thus avoiding the dependence on fixed simplified models in traditional methods.
2. **Symbolic Expressions**: Use symbolic expressions to seamlessly integrate the high - order driving - point functions of the RC network with the pre - characterized library of drivers, calculate the dynamic capacitance values, and consider their interdependencies at the same time.
3. **High Efficiency and Accuracy**: Compared with traditional methods, this algorithm can quickly and accurately predict the non - digital behavior of signal lines even for complex and large - scale RC loads.
4. **Scope of Application**: This algorithm performs well in calculating delay, power consumption and signal - line electromigration, etc., and has good resolution and efficiency.
### Specific technical details:
- **Basic Assumption**: It is assumed that the current (or voltage) response of the driver when loaded with different pure - capacitance loads is known. These data are usually provided by the foundry. If not available, they can be pre - characterized by SPICE simulation.
- **Current - Response Calculation**: Use the capacitance - current relationship \(I = C\cdot\frac{dV}{dt}\), calculate the current response of the RC network by interpolating and expanding the data in the driver response table.
- **Piecewise Processing**: Divide the output curve into multiple regions, determine the average capacitance load in each region, and generate the required \(I(t)\) or \(V(t)\) by interpolation or extrapolation.
- **Matching Strategy**: Use binary search combined with monotonicity strategy to quickly find the effective capacitance value \(C_{eff}\) at each step.
- **Symbolic Expressions**: Accelerate the calculation and eliminate the cumulative error through symbolic expressions, especially in large - scale or small - value RC networks.
### Experimental Verification:
The paper carried out experimental verification on a 40 - nm CMOS process. The results show that this algorithm is superior to traditional methods and SPICE gold - standard results in terms of resolution and efficiency, and is especially suitable for applications such as calculating delay, power consumption and signal - line electromigration.
### Conclusion:
The RC current - response algorithm based on dynamic capacitance matching (DCM) proposed in this paper is not only superior to traditional methods in terms of accuracy, but also performs well in terms of computational efficiency and configuration flexibility, and has broad application prospects.