[FEA] Story - Supporting Approximate Count Distinct

[ttnghia]

This issue tracks the dependencies for supporting approximate count distinct using HyperLogLog algorithm.

• Design doc (WIP): https://docs.google.com/document/d/1ezAXhH7GejuaNDvIyqMBYfmsSypcDWinl2_tmLJd-O4

[jrhemstad]

Given there was already sizable discussion in https://github.com/rapidsai/cudf/issues/10652, can we just use that issue? Or is there a different intent with this one?

[ttnghia]

That issue has been diluted. Here I'm going to add checked list item for the necessary PRs/features we are going to add so people can keep track, like this: https://github.com/rapidsai/cudf/issues/10186.

[revans2]

This is still wanted

[revans2]

This is still wanted

[github-actions[bot]]

This issue has been labeled inactive-30d due to no recent activity in the past 30 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed. This issue will be labeled inactive-90d if there is no activity in the next 60 days.

[GregoryKimball]

@etseidl FYI, our friends in Spark-RAPIDS are also interested in the HyperLogLog algorithm. 😄

[GregoryKimball]

Also see https://github.com/NVIDIA/cuCollections/pull/429

[vyasr]

Echoing Jake's question from two years ago: do we need this issue? Can we consolidate discussion in #10652? At this point it seems like we're just forced to post updates in two places. @ttnghia WDYT?

[res-life]

I'm working on this.

[res-life]

If no one has already started this, then let me pick this one.

[vyasr]

To avoid fragmentation, I'm going to close this as a dup of #10652 so that we can focus discussion there.