Merge pull request #350 from Exawind/small_problem_gpu_optimization #848

Job	Run time
Correctness-Linux (g++, Release, all)	1h 57m 3s
Correctness-Linux (g++, Release, none)	2h 3m 37s
Correctness-Linux (g++, Debug, all)	1h 54m 7s
Correctness-Linux (g++, Debug, none)	1h 57m 51s
Correctness-Linux (clang++, Release, all)	1h 57m 49s
Correctness-Linux (clang++, Release, none)	2h 1m 38s
Correctness-Linux (clang++, Debug, all)	1h 54m 19s
Correctness-Linux (clang++, Debug, none)	1h 55m 54s
	15h 42m 18s

Provide feedback