2 ms·
it's more about space efficiency. if a model can fit in only one or two cachelines, and get you "close enough", you may end up touching a lot less memory than i
by koverstreet 8y ago
it's more about space efficiency. if a model can fit in only one or two cachelines, and get you "close enough", you may end up touching a lot less memory than if you just did a binary search.