May 2022 Update

Documentation

Core Library

  • Add support for user-defined join operators.

  • Build abstract and portable SIMD using xsimd.

  • Add AlignedStlAllocator for folly F14 containers.

  • Add Substrait FilterNode and FieldAccessTypedExpr, CastTypedExpr support.

  • Add Decimal type.

  • Add Checkpointing for SsdCache.

  • Add SSE4.2 support.

  • Add TPC-H Connector.

  • Add support for constant partitioning columns.

  • Add fast path for const, flat encoded primitive arguments in SimpleFunctionAdapter.

  • Add support for more types in HivePartitionFunction.

  • Add support for Apple M1.

  • Add Velox to Substrait converter.

  • Add spilling to HashAggregation.

  • Add support for anti join with filter.

Presto Functions

Performance and Correctness

  • Add benchmark for simple aggregates.

  • Add benchmark for row writer.

  • Reuse memory more often when evaluating simple functions.

  • Reuse memory for aggregation results when possible.

  • Optimize row writer performance by around 20%.

  • Add string writer benchmark.

Debugging Experience

  • Add memory allocations to printPlanWithStats.

  • Print output type in PlanNode::toString.

  • Add config property to enable cpu usage tracking for expressions.

  • Add number of input and output batches statistics for operators.

  • Add number of processed batches, vectors for expressions.

Credits

Abhash Jain, Adam Simpkins, Amithash Prasad, Arpit Porwal, Behnam Robatmili, Chad Austin, Deepak Majeti, Ernesto Avilés Vázquez, Ezgi Çiçek, James Xu, Jialiang Tan, Jie1 Zhang, Jimmy Lu, Jing Zhu, John Reese, Karteek Murthy, Kevin Wilfong, Krishna Pai, Laith Sakka, MJ Deng, Masha Basmanova, Muir Manders, Orri Erling, Patrick Stuedi, Pedro Eugenio Rocha Pedreira, Pyre Bot Jr, Rui Mo, Sergey Pershin, TJ Yin, Wei He, Zhenyuan Zhao, artem.malyshev, rui-mo, usurai, xuedongluan, yeyuqiang, yingsu00