4 ms·
> It's not really fair to compare a custom implementation against the standard unordered_map implementations, which But he is comparing the standard unordered_
by flurrything 8y ago
> It's not really fair to compare a custom implementation against the standard unordered_map implementations, which
But he is comparing the standard unordered_map implementation to his own implementation of the standard unordered_map (same API).
- allenz 8y agoIt's the same API, but I highly doubt that his implementation is fully compliant with the C++ standard.
- flurrything 8y agoIt is, and so is boost::unordered_map, that's the whole point of the comparison. Just check the implementation out.
- allenz 8y agoSource on that claim? Here's a comment chain where the author admits that his library is noncompliant: https://probablydance.com/2017/02/26/i-wrote-the-fastest-hashtable/#comment-3027 https://probablydance.com/2017/02/26/i-wrote-the-fastest-has.... Also note the complete lack of a standards test suite. This library uses Robin Hood hashing for speed. As my original link explains, standard implementations use chained buckets so that users can cache lookups and to have better worst-case performance: https://news.ycombinator.com/item?id=9675964 https://news.ycombinator.com/item?id=9675964