Todos los libros

Profesional

Apps Acerca de Coach Iniciar sesión Empezar a leer

Quantitative Finance · Glosario

¿Qué es Vectorised query execution, late materialisation?

También llamado: vectorised query execution · late materialisation

Definition 5.4 Research, Data and Risk Platforms · Capítulo 5 — Time-Series Databases and the Alternatives

Vectorised query execution runs each operator of a query plan on a batch of values of a column at a time (a vector of a few thousand), so that the per-call overhead is paid per batch and the inner loops are tight loops over arrays. Late materialisation delays assembling result rows from columns until after the filters: operators pass positions of qualifying rows, and only the surviving rows’ other columns are ever read.

Row-at-a-time and vectorised execution of the same plan. The vectorised engine calls each operator once per batch of a column and passes positions instead of rows (late materialisation), so the other columns of rows that fail the filter are never read.
Figure 5.3. Row-at-a-time and vectorised execution of the same plan. The vectorised engine calls each operator once per batch of a column and passes positions instead of rows (late materialisation), so the other columns of rows that fail the filter are never read.
Leer en el capítulo →