Documentation
Rate This Document
Findability
Accuracy
Completeness
Readability
SubmitClose

Change Description

Starting from version 25.2.RC1, BoostKit has officially frozen component-level versioning. This means that individual component packages such as BoostKit-SRA_Recall_1.2.0.zip will retain their current version identifiers (1.2.0) in future BoostKit releases.

New Features

  • The KVecTurbo sub-library code is open-sourced and can be used only after being compiled. The source code is released on Gitee, and the code branch version is v1.1.0.
  • KVecTurbo supports datasets with over ten million vectors, having increased the upper limit on base library capacity. However, setting the vector count excessively high can lead to memory allocation failures due to system resource constraints.
  • The KBest and KScaNN sub-libraries are not updated in this release, and the code is not open-sourced. For details about how to obtain, install, and use BoostKit-SRA_Recall_1.2.0.zip, see the Kunpeng Recall Algorithm Library Developer Guide.
  • Added the KNewPfordelta sub-library. It is released as source code on Gitee (branch v1.1.0) and must be built from source before use.

Modified Features

None

Deleted Features

None

Parent topic: V25.2.RC1