Abstract:The feasibility-seeking approach provides a systematic scheme to manage and solve complex constraints for continuous problems, and we explore it for the floorplanning problems with increasingly heterogeneous constraints. The classic legality constraints can be formulated as the union of convex sets. However, the convergence of conventional projection-based algorithms is not guaranteed as the constrain sets are non-convex. In this work, we propose a resetting strategy to greatly eliminate the the divergence issue of the projection-based algorithm for the feasibility-seeking formulation. Furthermore, the superiorization methodology (SM), which lies between feasibility-seeking and constrained optimization, is firstly applied to floorplanning. The SM uses perturbations to steer the feasibility-seeking algorithm to a feasible solution with shorter total wirelength. The proposed flow is extendable to tackle various constraints and variants of floorplanning problems, e.g., floorplanning with I/O assignment problems. We have evaluated the proposed algorithm on the MCNC benchmarks. We can obtain legal floorplans only two times slower than the branch-and-bound method in its current prototype using MATLAB, with only 3% wirelength inferior to the optimal results. We evaluate the effectiveness of the flow by considering the constraints of I/O assignment, and our algorithm achieve 8% improvement on wirelength.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the floorplanning problem in integrated circuit (IC) design, especially when it comes to input/output (I/O) pin assignment. Floorplanning is a crucial stage in the VLSI (Very - Large - Scale Integration) physical design flow, which affects the quality of subsequent design stages. Specifically, this problem can be described as placing a set of rectangular modules within a given rectangular area so that there is no overlap between these modules and factors such as bus length, congestion, and temperature are minimized. However, when diverse constraint conditions are introduced, such as boundary conditions, non - overlapping conditions, and I/O assignment conditions, traditional floorplanning methods may find it difficult to effectively handle these problems.
### Main contributions of the paper:
1. **Modeling of the Feasibility - Seeking Problem (FSP)**:
- Model the floorplanning problem as a feasibility - seeking problem, that is, find a point that satisfies all constraints in a finite family of constraint sets. This method simplifies the problem and makes it easier to handle complex constraint conditions.
2. **Generalized Projection Method with Reset Strategy**:
- Propose a Generalized Projection Method with Reset Strategy (RMAP) to handle the complex constraint sets in the floorplanning problem. This reset strategy improves the initial convergence behavior and avoids the situation where the algorithm gets trapped in infeasible solutions or oscillates.
3. **Wire - Length Superiorization Method**:
- Apply the Superiorization Method (SM) to find a better feasible solution to make the bus length shorter. By introducing perturbations during the iteration process, the algorithm is guided towards a better feasible solution.
4. **Extension to the I/O Assignment Problem**:
- The proposed algorithm flow can be extended to handle the floorplanning problem including I/O assignment. Experimental results show that compared with the Branch - and - Bound (B&B) method, this method achieves an 8% improvement in bus length while the time cost is within an acceptable range.
### Experimental Results:
- **Floorplanning without I/O Assignment**:
- On the MCNC benchmark set, the half - perimeter wire length (HPWL) of the proposed method (Per - RMAP) is only 3% longer than the optimal result, and the execution time is approximately 1.95 times that of the B&B method.
- **Floorplanning with I/O Assignment**:
- In the case of considering I/O assignment, the HPWL of the Per - RMAP method is 8% shorter than that of the B&B method, but the execution time increases by approximately 4.99 times.
### Conclusions and Future Work:
- This paper successfully overcomes the limitations of traditional methods in handling complex constraint conditions by modeling the floorplanning problem as a feasibility - seeking problem and combining the reset strategy and the superiorization method.
- Future work will focus on expanding the scalability of this method, verifying its performance by adding experiments on larger instances, and exploring its ability to handle more complex practical constraints (such as 2.5D floorplanning).
### Examples of Formulas:
- Calculation formula for half - perimeter wire length (HPWL):
\[
\text{HPWL}(x, y) := \sum_{e \in E} \left( \max_{p, q \in e} |x_{\text{pin}}^p - x_{\text{pin}}^q| + \max_{p, q \in e} |y_{\text{pin}}^p - y_{\text{pin}}^q| \right)
\]
- Calculation formula for the preference ratio in the projection method:
\[
\eta_k =
\begin{cases}
-\infty, & \text{if } c_k > T \text{ (and reset } c_k = 0 \text{)} \\
-\|z - P_k(z)\|, & \text{otherwise (and set } c_k = c_k + 1 \text{)}
\end{cases}
\]
Hope this information is helpful for you to understand the content of this paper!