I have used google benchmark library on Harris corner detection kernel (from polymage benchmarks). I want to know if the way I have used the google benchmark library here is correct or there can be a better way to do this.
Also, I want to know your comments on using sha1sum (or some library or implementing it manually) with google benchmark library to check for correctness of output. In case sha1sum (or any other checksum method) should not be used then what can I do to check for output when the output do not follow any general pattern and reference output file is large. Google Test can be a good option but I cannot find any kernel in test suite which uses google test.