datafusion-comet icon indicating copy to clipboard operation
datafusion-comet copied to clipboard

Is there any performance report in tpc-h or tpc-ds with Apache Spark and Gluten?

Open thexiay opened this issue 1 year ago • 3 comments

What is the problem the feature request solves?

No response

Describe the potential solution

No response

Additional context

No response

thexiay avatar Mar 04 '24 08:03 thexiay

@thexiay Not yet. We still have some issues to resolve until we can fully run TPC-H or TPC-DS and publish our benchmark results. Please stay tuned.

sunchao avatar Mar 04 '24 17:03 sunchao

@thexiay Not yet. We still have some issues to resolve until we can fully run TPC-H or TPC-DS and publish our benchmark results. Please stay tuned.

maybe you can upload One-click TPC-H or TPC-DS test script in other branch or main(e.g. you mentioned it in #141) and I can solve part of bugs while runing TPC-H or TPC-DS @sunchao

thexiay avatar Mar 06 '24 05:03 thexiay

@thexiay it's possible to run TPC-DS locally with Comet via following the steps in this file. #141 will require some work to implement.

sunchao avatar Mar 06 '24 05:03 sunchao

We now have the Comet Benchmarking Guide which shows current benchmark results, and all of the TPC-H scripts are open source as part of the new DataFusion Benchmarks repo, so I think we can close this issue now.

I know this issue asked about Gluten, but we do not plan on publishing benchmarks for other accelerators but the community is welcome to do so and we have provided all the scripts to enable this.

andygrove avatar Jun 06 '24 16:06 andygrove