Change 1713c6b (#24) removed the use of the ItemsCount counter because the resulting numbers didn't make much sense. So it was unreasonable to use the haystack size.
Every benchmark iteration searches for a single element, so it appears that the correct count to use is 1. The new throughput numbers now make more sense with this change.
Change 1713c6b (#24) removed the use of the
ItemsCount
counter because the resulting numbers didn't make much sense. So it was unreasonable to use the haystack size.Every benchmark iteration searches for a single element, so it appears that the correct count to use is 1. The new throughput numbers now make more sense with this change.
Results: