Abstract:Poisson's equation has been used in VLSI global placement for describing the potential field caused by a given charge density distribution. Unlike previous global placement methods that solve Poisson's equation numerically, in this paper, we provide an analytical solution of the equation to calculate the potential energy of an electrostatic system. The analytical solution is derived based on the separation of variables method and an exact density function to model the block distribution in the placement region, which is an infinite series and converges absolutely. Using the analytical solution, we give a fast computation scheme of Poisson's equation and develop an effective and efficient global placement algorithm called Pplace. Experimental results show that our Pplace achieves smaller placement wirelength than ePlace and NTUplace3. With the pervasive applications of Poisson's equation in scientific fields, in particular, our effective, efficient, and robust computation scheme for its analytical solution can provide substantial impacts on these fields.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is how to calculate the potential field more accurately by analytically solving the Poisson equation in VLSI global placement, thereby optimizing the placement of circuit blocks and reducing the wire length. Specifically, the paper proposes an analytical solution method to replace the existing numerical solution methods in order to improve the quality and efficiency of the solution.
### Background and Motivation
In VLSI physical design, the placement problem is to place circuit blocks in a fixed chip area so that the blocks do not overlap with each other and optimize certain cost metrics (such as wire length). As the technology node enters the deep - nanometer era, the circuit design integrating billions of transistors has increasingly high requirements for the performance of the placement engine. Many placement algorithms have been developed, mainly including annealing - based methods, partitioning - based methods, and analysis - based methods. Among them, analysis - based methods can usually obtain the best placement quality and have good scalability.
### Application of Poisson Equation in VLSI Placement
The Poisson equation is used to describe the potential field caused by a given charge density distribution in VLSI global placement. Some existing placement methods solve the Poisson equation by numerical methods. For example, ePlace uses the fast Fourier transform (FFT) to quickly calculate the potential field distribution, so as to quickly disperse circuit blocks while minimizing the wire length.
### Main Contributions of the Paper
1. **Analytical Solution Method**: The paper directly solves the Poisson equation and obtains an analytical solution in the form of an infinite series, and proves the absolute convergence of this infinite series.
2. **Fast Calculation Scheme**: Based on the analytical solution, the paper proposes a fast calculation scheme for calculating the Poisson equation in \(O(n \log n)\) time, where \(n\) is the number of circuit blocks. Then, an efficient and effective global placement algorithm Pplace is developed.
3. **Experimental Results**: Through comparison with two leading placers, ePlace and NTUplace3, on the ISPD 2005 and ISPD 2006 benchmark test sets, the experimental results show that Pplace is superior to ePlace in wire length optimization. By replacing the density control method in NTUplace3 and further embedding the fast calculation of the Poisson equation, the total wire length is reduced by 11%, which is an unprecedented and significant improvement.
4. **Widespread Impact**: Due to the wide application of the Poisson equation in multiple scientific fields, the efficient, effective, and robust analytical solution calculation scheme proposed in this paper has an important impact on these fields.
### Specific Steps of the Analytical Solution Method
1. **Analytical Solution of the Poisson Equation**:
- Expand the density function into an infinite series form by the method of separation of variables.
- Solve the analytical expression of the potential function \(\psi(\hat{x}, \hat{y})\), in the form of:
\[
\psi(\hat{x}, \hat{y})=\sum_{u = 0}^{\infty}\sum_{p = 0}^{\infty}a_{u,p}\cos\left(\frac{u\pi\hat{x}}{W}\right)\cos\left(\frac{p\pi\hat{y}}{H}\right)
\]
- Calculate the coefficient \(a_{u,p}\) by integration.
2. **Fast Calculation Scheme**:
- Divide the chip area into a uniform grid of \(m\times m\).
- Quickly calculate the coefficient matrix \(a_{u,p}\) by the fast Fourier transform (FFT).
- Use the inverse fast Fourier transform (IFFT) to calculate the potential value \(\psi(l, j)\) and the electric field value \(\xi(l, j)\) at each grid point.
### Experimental Results
- **Wire Length Optimization**: Pplace is superior to ePlace in wire length optimization.
- **Running Time**: The running time of Pplace is comparable to that of ePlace, but it performs better in wire length.
- **Robustness**: By replacing the density control method in NTUplace3, the robustness and effectiveness of Pplace are further verified.
### Conclusion
The paper provides a new global placement algorithm Pplac by analytically solving the Poisson equation.