experimental_cagra_index¶

The experimental_cagra_index system variable controls whether the CAGRA (Cuda ANN Graph) vector index plugin is enabled. When set to 0 (default), CREATE INDEX ... USING cagra statements are rejected. Set it to 1 before creating or rebuilding a CAGRA index. CAGRA is a GPU-accelerated graph-based ANN algorithm and requires GPU hardware for index construction. This is an experimental feature gated by a session-level flag.

The experimental_cagra_index system variable gates access to the CAGRA vector index algorithm in MatrixOne. CAGRA is a GPU-accelerated graph-based algorithm for approximate nearest neighbor search, providing high recall and performance at scale. Index construction requires a GPU. Because this is an experimental feature, the variable defaults to 0 and must be explicitly enabled before use.

Syntax¶

SET GLOBAL experimental_cagra_index = 0; SET SESSION experimental_cagra_index = 1;

Query the current value:

SHOW VARIABLES LIKE ‘experimental_cagra_index’; SELECT @@experimental_cagra_index; SELECT @@global.experimental_cagra_index;

Arguments¶

Property	Value
Variable Type	`bool`
Scope	`Both`
Dynamic	`Yes`
Default Value	`0`
Optional Value	`0`, `1`

Examples¶

DROP DATABASE IF EXISTS cagra_demo;
CREATE DATABASE cagra_demo;
USE cagra_demo;

SET experimental_cagra_index = 1;

SELECT @@experimental_cagra_index;
SELECT @@global.experimental_cagra_index;

SET experimental_cagra_index = 0;

DROP DATABASE cagra_demo;

Constraints¶

experimental_cagra_index defaults to 0. Attempting CREATE INDEX ... USING cagra while the variable is 0 causes the operation to fail.
CAGRA requires GPU hardware for index construction. It will not function on CPU-only builds of MatrixOne.
This is an experimental feature. Its behavior, performance characteristics, and API may change in future releases.