A Fast x86 Implementation of Select · HackerPapers discussions

A Fast x86 Implementation of Select — mentioned by judofyr

Thu, 06 Mar 2025 19:19:49 GMT

I can speak a bit about one of the approaches: "Practical Entropy-Compressed Rank/Select Dictionary" by Daisuke Okanohara and Kunihiko Sadakane. This presents two different variants: One for dense (i.e. more than 50% of the bits are set) and one for sparse. The dense implementation is basically based around partitioning them into "blocks" of a given size and then you can find the block by doing `block[idx / block_size]`. It then also groups each block into sub-blocks which helps you even further. All of these are additional data structures (very much like an index) which are stored next to…

Paper and discussion history. A mention is not an endorsement.

A Fast x86 Implementation of Select — mentioned by Bayart

Fri, 11 Jun 2021 20:32:09 GMT

The paper, if anyone wants to save some clicks : https://arxiv.org/abs/1706.00990

Paper and discussion history. A mention is not an endorsement.

A Fast x86 Implementation of Select — mentioned by dgacmu

Sun, 24 May 2020 18:33:23 GMT

This paper was first published in 2008 (and has since been beaten several times over by more modern techniques). a quick glance doesn't suggest that there's anything obviously new in this version, is there a reason it has a new date? Original: https://dl.acm.org/doi/10.5555/1788888.1788900 to those curious about this area, one of the most recent and fastest results is this one from pandey et al: https://arxiv.org/abs/1706.00990

Paper and discussion history. A mention is not an endorsement.

A Fast x86 Implementation of Select — HN submission

Sat, 13 Oct 2018 23:44:53 GMT

A Fast x86 Implementation of Select (2017) [pdf]

90 points; 18 discussion replies on HN. Paper and source history. A submission is not an endorsement.

A Fast x86 Implementation of Select — HN submission

Wed, 07 Jun 2017 08:21:18 GMT

A Fast X86 Implementation of Select

3 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.