Expand description
DataChunk - the fundamental unit of vectorized execution.
A DataChunk holds a batch of rows in columnar format. Processing data in batches (typically 1024-2048 rows) lets the CPU stay busy and enables SIMD.
Structs§
- Data
Chunk - A batch of rows stored column-wise for vectorized processing.
- Data
Chunk Builder - Builder for creating DataChunks row by row.
Constants§
- DEFAULT_
CHUNK_ SIZE - Default chunk size (number of tuples).