NVIDIA / NVIDIA/cudf

[FEA] Support segmented reductions (MIN/MAX/COUNT DISTINCT) in cuDF `list` accessor

Open
#12,301 13 comments 0 reactions 0 assignees View on GitHub
0 - Backlog feature request Python
Dominant language
C++
Stars
9.8k
Forks
1.1k
Avg merge
3d 6m
Merged PRs (30d)
278

Description

**Is your feature request related to a problem? Please describe.**

Currently, we use cuDF to implement distributed group aggregation operations. That is, GROUP BY + DISTINCT is performed on the local node, and GROUP BY + MIN/MAX/COUNT DISTINCT are performed on the merge node. The LIST is transferred as an intermediate format, but the aggregation operation cannot be implemented.

**Describe the solution you'd like**

Group + Aggregation supports the list format.

![捕获2](https://user-images.githubusercontent.com/43532055/205540928-0359139e-03ea-4c71-bd8d-0daaf0d5f875.PNG)

**Describe alternatives you've considered**

Aggregation operators are supported in the list format. For example, the current list supports only `len`, and aggregation operators such as `count`, `min`, `max`, and `avg` are expected to be added.

![捕获](https://user-images.githubusercontent.com/43532055/205540951-3d8b4f09-7227-4cb1-836e-49e79c428e31.PNG)

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.