Bulk qualification, the process of evaluating single-table predicates on multiple column values in parallel, is now applied to these SQL DML request steps:
- Retrieve, sum, and join steps on these tables:
- Primary index row
- No primary index row
- Row spool
- Sum and join steps on these tables:
- Columnar primary AMP index
- Columnar no primary index
- Columnar primary index
The retrieve step was supported in Release 15.10.
Bulk qualification uses single-instruction multiple-data (SIMD) parallel processing.
Bulk qualification may not be enabled in a step, based on restrictions. To detect whether bulk qualification was enabled in a step, users can check the InMemBulkQualStatus field of the QryLogStepsV view.
Benefits
This feature improves the CPU performance of retrieve, join, and sum steps.
Considerations
This feature is enabled only if In-Memory Optimization is enabled. In-Memory Optimization is disabled by default. To enable it, contact your Teradata representative.
Additional Information
For more information on the InMemBulkQualStatus field, see Teradata Vantage™ Data Dictionary, B035-1092.