Ferdinand Xu

Intel 大数据团队招聘

4

有兴趣的童鞋，请砸简历到 [[email protected]](mailto:[email protected]) , 如果合适，我会帮大家内推, 详情~ JD请参考如下链接: - https://intel.wd1.myworkdayjobs.com/External/job/PRC-Shanghai/Big-Data-Analytics-Software-Development-Engineer_JR0209066 - https://intel.wd1.myworkdayjobs.com/External/job/PRC-Shanghai/Big-Data-Analytics-Software-Development-Engineer_JR0209065-1 - https://intel.wd1.myworkdayjobs.com/External/job/PRC-Shanghai/Deep-learning-Architect_JR0172661-1 - https://intel.wd1.myworkdayjobs.com/External/job/PRC-Shanghai/AI-Frameworks-Engineer_JR0203146

PARQUET-41: Add bloom filters to parquet statistics

3

It's the PR in mr part.

This fixes [CRYPTO-2: Create benchmark for Apache Commons Crypto](https://issues.apache.org/jira/browse/CRYPTO-2).

[FEA] Support short-circuit evaluation for expensive expression like rlike

1

**Is your feature request related to a problem? Please describe.** ```Rlike``` was an very expensive [cudf operation](https://github.com/rapidsai/cudf/blob/f9ac4277f50163a7da2006460034aa3e45c8744e/cpp/src/strings/contains.cu#L59) compared to other operator like compare. One option is mentioned in https://github.com/NVIDIA/spark-rapids/issues/10600 that...

[FEA] Optimize Shuffle coalesce performance when handling very small sized partitioned batches

1

**Is your feature request related to a problem? Please describe.** Currently shuffle will partition the columnar batch into partitioned batches. https://github.com/NVIDIA/spark-rapids/blob/65ec25d85503484874d3b157d4e2775159187436/sql-plugin/src/main/scala/com/nvidia/spark/rapids/GpuHashPartitioningBase.scala#L35-L47 And those partitioned batches will go through ```GpuColumnarBatchSerializer``` in...