Skip to main content

Table 10 Summary of CUDAlign stages, including the SW phase it belongs to and the processor where it is executed

From: DNA sequences alignment in multi-GPUs: acceleration and energy payoff

GPU model

Stage 1

Stage 2

Stage 3

Total

Versus GTX 980

 
 

Average power (watts)

 

GCUPS/W

GTX 680 (Kepler)

133.10 W

119.13 W

79.16 W

132.63 W

+28%

0.57

GTX 980 (Maxwell)

102.95 W

114.44 W

81.27 W

103.06 W

 

1.09

Titan X Maxwell

189.45 W

172.71 W

115.48 W

189.00 W

+83%

0.84

Titan X Pascal

194.54 W

176.52 W

112.56 W

193.96 W

+88%

1.42

 

Execution time (seconds)

 

GCUPS

GTX 680 (Kepler)

33207.27 s

642.37 s

128.70 s

33978.34 s

+50%

75.10

GTX 980 (Maxwell)

22302.24 s

291.50 s

46.65 s

22640.40 s

 

112.71

Titan X Maxwell

15870.33 s

222.11 s

46.51 s

16138.95 s

- 29%

158.10

Titan X Pascal

9114.21 s

144.75 s

33.02 s

9291.98 s

- 59%

274.62

 

Energy consumption (kilojoules)

 

Cost

GTX 680 (Kepler)

4419.88 kJ

76.52 kJ

10.18 kJ

4506.58 kJ

+93%

0.1627 €

GTX 980 (Maxwell)

2296.22 kJ

33.36 kJ

3.79 kJ

2333.37 kJ

 

0.0842 €

Titan X Maxwell

3006.63 kJ

38.36 kJ

5.37 kJ

3050.36 kJ

+30%

0.1101 €

Titan X Pascal

1773.07 kJ

25.55 kJ

3.71 kJ

1802.33 kJ

- 23%

0.0650 €

  1. The comparison on sixth column takes GTX 980 as baseline