liliu-z
liliu-z
From `Knowhere` side, we can consider force setting omp thread num in `Wrapper` layer to ensure the consistency performance between different kinds of indexes. This num can be controlled by...
We only support **IP** for now, and **Cosine** is not supported yet. **IP** without normalization is workable but with super low recall (it doesn't make sense in Mathematics). **IP** +...
@cydrain Can you help do a further check on why this happens? Appreciate it!
> so all graph with IP should have similar problem? build the graph with pre cluster might help on it To my understanding, when we use IP in HNSW, the...
It looks like a right place to optimize /assign @congqixia for further check
Did you try cases in VDBBench with filtering? Curios about the potential improvement
Try enlarge `max_empty_result_buckets` according to https://milvus.io/docs/index.md#SCANN