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

Table 3

Comparison of PCSRI and PCSRII without communication on two GPUs.

Matrix	ET (GPU)	PCSRI (2 GPUs)			PCSRII (2 GPUs)
		ET	SD	PE	ET	SD	PE
2cubes_sphere	0.4444	0.2670	0.0178	83.21	0.2640	0.0156	84.17
scircuit	0.3484	0.2413	0.0322	72.20	0.2250	0.0207	77.41
Ga41As41H72	4.2387	2.3084	0.0446	91.81	2.3018	0.0432	92.07
F1	6.5544	3.8865	0.7012	84.32	3.5710	0.2484	91.77
ASIC_680ks	0.8196	0.4567	0.0126	89.72	0.4566	0.0021	89.74
ecology2	1.2321	0.6665	0.0140	92.42	0.6654	0.0152	92.58
Hamrle3	1.7684	0.9651	0.0478	91.61	0.9208		96.02
thermal2	2.0708	1.0559	0.0056	98.06	1.0558	0.0045	98.06
cage14	5.9177	3.4757	0.5417	85.13	3.1548	0.0458	93.78
Transport	4.7305	2.4665	0.0391	95.89	2.4655	0.0407	95.93
G3_circuit	1.9731	1.0485	0.0364	94.08	1.1061	0.1148	89.18
kkt_power	4.3465	2.7916	0.7454	77.85	2.2252	0.0439	97.66
CurlCurl_4	5.1605	2.7107	0.0347	95.18	2.7075	0.0244	95.30
memchip	3.8257	2.1905	0.3393	87.32	2.0975	0.2175	91.19
Freescale1	5.0524	3.0235	0.5719	83.55	2.8175	0.2811	89.66