Weekly BigData & ML Roundup – Mar. 23, 2017

LLVM for Data Processing When Apache Spark launched Tungsten, there was a hint of incorporating LLVM into data processing pipeline to make use of modern CPU features and its superb machine code for performance boost. LLVM appears again with Weld, a new code generation project for data analytics. The project claims that Tensorflow, Spark, and Numpy can be accelerated up to 30x with just few operations of Weld! Interestingly, Matei Zaharia is in its contributor list. ETL Starter Kit Extract, Transform, Load (ETL) refers to a process in database usage and especially in data warehousing.