mFAR: Multilevel Fixed-Points Addition-Based VLSI Placement

DPlace: Anchor Cell-Based Quadratic Placement with Linear Objective

“…The fixed point based approach adds artificial pins and nets to move cells. mFar [9] uses multiple fixed virtual pins for each cell in every iteration, one is used to maintain a cell's force equilibrium state, and others are applied to perturb the cell. FastPlace [11] uses one fixed virtual pin for both purposes.…”

“…simulated annealing [3], iterative partitioning based approach [4,5,6], and analytical placement approach [7,8,9,10,11,12,13,14].…”

“…Since HPWL model is not smooth and derivable, quadratic placement optimizes the quadratic form of HPWL [7,8,9,10,11,12,14], and nonlinear model placement [16] [13] [17] adopts a nonlinear estimation of HPWL model, such as the log-sum-exponential wire length approximation patented by Naylor et al [18].…”

FastPlace 3.0: A Fast Multilevel Quadratic Placement Algorithm with Placement Congestion Control

“…Existing placement algorithms employ various approaches including simulated annealing [24,25], partitioning [1,2,7,29] and analytical placement [4,[9][10][11]16,17,21,27,28]. Analytical placement algorithms based on the quadratic objective funtion (also called quadratic placers) are very popular as they are quite efficient and also give good quality of results.…”

“…Analytical placement algorithms based on the quadratic objective funtion (also called quadratic placers) are very popular as they are quite efficient and also give good quality of results. They typically employ a flat placement methodology [9][10][11]17,27,28] so as to maintain a global view of the placement problem.…”