Closed xiaguan closed 7 months ago
For tpch 10gb q1 , which has a lot of aggregation. 1:10 -> 40s It's not a pull request indeed. I think we need a paralleism executor.
Good experiment! But I think the final solution should be building the same executors for each core. The executor itself should focus on processing data in a sequential manner.
For tpch 10gb q1 , which has a lot of aggregation. 1:10 -> 40s It's not a pull request indeed. I think we need a paralleism executor.