CUDAMPF: a multi-tiered parallel framework for accelerating protein sequence search in HMMER on CUDA-enabled GPU

BMC Bioinformatics

Table 2 Benchmark of elapsed time (ms) for each steps of runtime compilation

Steps of Runtime	MSV/SSV kernel		P7Viterbi kernel
Compliation	Intel i5^a	Intel Xeon^b	Intel i5	Intel Xeon
Read.cu.cuh file into string	0.05/0.04	0.07/0.08	0.05	0.09
Create and compile nvrtcProgram	450/431	855/812	422	836
Get PTX string and kernel handle	0.35/0.77	5.38/5.63	0.27	7.64

ISSN: 1471-2105