A New Length-Based Algebraic Multigrid Clustering Algorithm

Table 3

Placement wire length and runtime results comparing one- and two-level AMG-LE clustering with length-driven unclustering to baseline placements without using clustering with three placers on the ICCAD04 benchmark suite.

(a) HPWL

Circuit	Capo			Fastplace			mPL
		AMG-LE			AMG-LE			AMG-LE
	Baseline (×10^5)	1Lvl (%)	2Lvl (%)	Baseline (×10^5)	1Lvl (%)	2Lvl (%)	Baseline (×10^5)	1Lvl (%)	2Lvl (%)
ibm01	25	3.1	6.4	24	−1.8	1.3	24	4.6	4.2
ibm02	51	4.4	5.4	54	3.1	4.6	52	1.3	1.1
ibm03	77	1.8	9.8	80	5.1	6.8	82	−7.0	−3.6
ibm04	93	14.8	15.7	86	7.9	7.7	109	−5.7	−4.4
ibm05	103	1.1	2.1	101	0.7	0.5	93	−5.2	−5.1
ibm06	67	0.6	3.8	99	38.1	33.9	88	5.8	2.7
ibm07	126	9.3	11.9	123	7.4	12.2	124	−3.1	−4.0
ibm08	137	7.9	8.1	147	−4.1	10.1	211	4.7	4.2
ibm09	145	4.4	6.6	155	8.6	8.0	189	2.9	1.7
ibm10	318	4.0	4.9	362	10.8	12.2	363	2.0	1.8
ibm11	210	4.1	7.0	225	11.0	10.1	243	−0.1	−2.6
ibm12	413	10.9	14.5	410	11.4	14.4	461	−1.1	1.1
ibm13	266	3.4	7.6	273	12.1	11.8	324	−0.4	1.1
ibm14	392	2.6	4.0	478	14.5	21.7	824	6.5	6.6
ibm15	544	6.1	5.6	577	8.6	12.2	1001	−2.9	−5.0
ibm16	629	5.6	6.4	680	11.5	11.5	931	6.7	5.7
ibm17	737	2.7	3.8	810	9.3	12.2	1144	7.5	7.2
ibm18	458	2.1	3.3	574	19.0	19.8	885	11.0	10.7
Average	—	4.9	7.1	—	9.6	11.7	—	1.5	1.3

(b) Runtime

Circuit	Capo			Fastplace			mPL
		AMG-LE			AMG-LE			AMG-LE
	Baseline (s)	1Lvl (%)	2Lvl (%)	Baseline (s)	1Lvl (%)	2Lvl (%)	Baseline (s)	1Lvl (%)	2Lvl (%)
ibm01	193	−3	−75	26	−5	−62	109	−69	−152
ibm02	329	−24	−112	46	−14	−72	236	−91	−171
ibm03	487	−21	−105	46	−20	−82	221	−108	−210
ibm04	512	−23	−110	51	−17	−90	250	−87	−195
ibm05	411	−25	−124	36	−35	−81	174	−118	−183
ibm06	580	−19	−115	96	29	−6	390	−83	−197
ibm07	920	−21	−105	96	−9	−82	387	−100	−198
ibm08	941	−20	−121	99	−31	−158	1109	−105	−197
ibm09	1198	−29	−119	122	−6	−66	898	−136	−221
ibm10	1868	−12	−120	286	4	−62	1450	−126	−210
ibm11	1834	−30	−121	135	−66	−139	1084	−84	−193
ibm12	1996	−22	−124	282	−12	−52	1517	−115	−222
ibm13	2387	−16	−120	239	−43	−118	1235	−110	−218
ibm14	3327	−29	−129	530	−42	−121	2189	−93	−213
ibm15	5781	−30	−142	624	−57	−197	3969	−101	−247
ibm16	4917	−35	−132	722	−75	−217	5457	−57	−118
ibm17	5286	−31	−135	1133	−33	−101	2788	−131	−228
ibm18	4215	−39	−142	1589	3	−57	3106	−125	−257
Average	—	−24	−119	—	−24	−98	—	−102	−202