The kfs program reads in multiline FASTA records, counts canonical kmers using Emil Ernerfeldt's emilib::HashMap hash table, and measures time taken to read in and process records. A discussion about performance characteristics compared with the C++ STL std::unordered_map is available from the author.
$ make kfs
Specify variables K (integer) and FASTA (path to FASTA sequences).
$ /usr/bin/time -l ./kfs -k ${K} -i ${FASTA}
...