An Enhanced Deep Reinforcement Learning-Based Global Router for VLSI Design

Table 3

Comparison of wire length and overflow of two routers.

No.	Wire length (WL)		Overflow (OF)		Optimization rate (%)
	-based	DDQN-based	-based	DDQN-based	WL	OF
1	163	165	27	0	-1.23	100
2	159	165	19	0	-3.77	100
3	243	248	16	0	-2.06	100
4	249	245	12	0	1.60	100
5	272	274	22	0	-0.74	100
6	277	290	32	0	-4.69	100
7	464	466	8	0	-0.43	100
8	496	504	19	0	-1.61	100
9	932	934	16	0	-0.21	100
10	787	794	14	0	-0.89	100
Average					-1.40	100