NVIDIA / NVIDIA/cudf

[FEA] Implement parallel reading of Parquet footers without using the additional thread pool

Open
#18,082 0 comments 2 reactions 1 assignee Claimed by @vuule View on GitHub
cuIO feature request libcudf
Dominant language
C++
Stars
9.8k
Forks
1.1k
Avg merge
3d 6m
Merged PRs (30d)
278

Description

https://github.com/rapidsai/cudf/pull/17957 optimizes ingest of multiple Parquet files, using an internal thread pool to issue read requests. Parallelism of the read operations is already implemented in kvikIO, so we should not need to launch these reads from multiple threads.

With an [async read interface](https://github.com/rapidsai/cudf/pull/18018), we could read all footers in a loop from a single thread.

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.