We really should update that FAQ. Drill could, but does not yet, take advantage of SIMD instructions. Drill is columnar, which means that the in memory representation of data is columnar and everything in Drill operates on columns of data instead of row by row operation. So, potentially we could use SIMD instructions to to compute values on entire arrays of data but we do not do that. Some of the work done in Drill has now moved to Apache Arrow and the hope is that we will incorporate SIMD operations in Arrow and then merge back into Drill. You might want to take a look at Arrow.
On Wed, Jul 6, 2016 at 6:00 AM, Praveen Krishna < [email protected]> wrote: > In one of the FAQ for Apache Drill it states that > > > "Drill takes advantage of the latest SIMD instructions available in modern > processors." > > > I am curious to know how SIMD vectorization is implemented Drill. >
