A Novel CSR-Based Sparse Matrix-Vector Multiplication on GPUs

Table 5

Comparison of PCSRI and PCSRII with communication on two GPUs.

Matrix	ET (GPU)	PCSRI (2 GPUs)			PCSRII (2 GPUs)
		ET	SD	PE	ET	SD	PE
2cubes_sphere	0.4444	0.2494		89.09	0.2503		88.75
scircuit	0.3484	0.2234	0.0154	77.95	0.2165	0.0070	80.44
Ga41As41H72	4.2387	2.3516	0.0030	90.12	2.3795	0.0521	89.07
F1	6.5544	3.9252	0.6948	83.49	3.6076	0.2392	90.84
ASIC_680ks	0.8196	0.4890	0.0113	83.80	0.4998	0.0178	81.99
ecology2	1.2321	0.6865		89.74	0.6863		89.76
Hamrle3	1.7684	1.0221	0.0209	86.50	1.0066	0.0170	87.84
thermal2	2.0708	1.1403	0.0230	90.80	1.1402	0.0203	90.81
cage14	5.9177	3.5756	0.5644	82.75	3.2244	0.0196	91.76
Transport	4.7305	2.4623	0.0203	96.05	2.4550	0.0183	96.34
G3_circuit	1.9731	1.1215	0.0189	87.96	1.1766	0.0896	83.84
kkt_power	4.3465	2.9539	0.6973	73.57	2.4459	0.0356	88.85
CurlCurl_4	5.1605	2.7064	0.0092	95.34	2.7049		95.39
memchip	3.8257	2.3218	0.3467	82.39	2.2243	0.1973	85.99
Freescale1	5.0524	3.1216	0.5868	80.92	2.9367	0.3199	86.02