But the collected test data could be much more useful to me and maybe other users, when you are able to restrict datasets to include e.g. only results for §test_x when the architecture is not a multi core, or e.g. is not more powerful than 2 GHz, etc. So you can compare it more easily with own tests.