Skip to main content

Table 3 Performance comparison between CUDASW++ 1.0, CUDASW++ 2.0 and NCBI-BLAST

From: CUDASW++2.0: enhanced Smith-Waterman protein database search on CUDA-enabled GPUs based on SIMT and virtualized SIMD abstractions

Software

Performance

 

Time(h)

GCUPS

Optimized SIMT (BL62, 10-2 k)

8.00

28.8

Partitioned (BL62, 10-2 k)

11.15

20.7

Partitioned (BL50, 10-3 k)

11.71

19.7

NCBI-BLAST(BL62, 10-2 k)

9.56

24.1

NCBI-BLAST(BL50, 10-3 k)

51.45

4.5

CUDASW++ 1.0 (BL62, 10-2 k)

14.12

16.3