Release Announcement: v1.4.0

We will release v1.4.0 this week.

From v1.3, Vald has some updating!

The main update is here:

LinearSearch of NGT

We have added the new API, LinearSearch, since NGT has released Linear Search CAPI from v1.13.8.

You can use LinearSearchas a simple search method or tool for tuning ANN search parameters.

We will post “HOW TO USE LinearSearch” as another blog post.

Support FP16 mode for NGT

NGT added the new features from v1.14.0.

FP16 mode improves the memory usage of NGT decreases than FP32 but may reduce search accuracy a bit.

If you’d like to reduce memory usage, you can try it!

You can use it by setting float16 at agent.ngt.object_type.

Pyroscope with Vald cluster

To more efficiently find the problems in your Vald cluster, continuous profiling is a good solution, so we have introduced Pyroscope.

You can use it by following the steps.

The sample image of Vald Cluster

We will also post focus on the story in another blog post 🙂

Added original gRPC status codec

We have implemented the original gRPC custom codec.

If your Vald cluster version is mismatched to the Vald client library version, it may not work well.

Improved search config and error code

From user feedback, we have decided to add a new parameter min_num at SearchConfig and a new gRPC status DEADLINE_EXCEEDEDfor an easy-to-predict error log.

The detail changes are below:

Follow Kubernetes v1.23

We also follow the new features of Kubernetes v1.23.

For more information about this change, please refer to below.

That’s all of the main updates.

For more details, please refer to below when v1.4 is available.

If you have any questions or any requests, please feel free to contact us 🙂

We are happy to your feedback!

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
vald.vdaas.org

A highly scalable distributed fast approximate nearest neighbor dense vector search engine.