I wonder if it is possible to develop a CUDA application that will search for given patterns in a text corpus (similar to grep in linux). The patterns will be loaded once in device memory at startup. Then, the data of each given file will be copied to the device where the string search will take place. The results will be copied back to host memory.
a) Is CUDA suitable for this kind of applications ?
b) Does anybody know if there are any good SIMD algorithms suitable for this job?
c) Do you know if something similar exists ?
Thanks in advance!