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

Table 2

Comparison of wire length and congestion standard deviation of two reward functions.

No.	Wire length (WL)		Congestion standard deviation (Std)		Optimization rate (%)
	Function 1	Function 2	Function 1	Function 2	WL	Std
1	165	165	0.95	0.85	0.00	10.53
2	168	165	0.94	0.94	1.79	0.00
3	257	248	1.30	1.12	3.50	7.44
4	249	245	1.23	1.16	0.40	5.69
5	277	274	1.31	1.23	1.08	2.38
6	293	290	1.32	1.23	1.02	6.82
7	467	466	0.83	0.81	0.21	4.71
8	521	504	0.95	0.89	3.26	6.32
9	937	934	1.36	1.29	0.32	7.19
10	801	794	1.16	1.09	0.87	6.03
Average					1.25	5.71