Is this your first time submitting a feature request?
[X] I have searched the existing issues, and I could not find an existing issue for this feature
[x] I am requesting a straightforward extension of existing functionality
Describe the feature
In the Pincone 2.x.x SDK, we have the ability to pass a batch of vectors to index.query method. In the Pinecone 3 SDK, we can only pass a single vector to index.query. We need high performance on the query method because we want to batch process a large vector dataset.
It makes it very slow for reading operations. I found that the Pinecone 2 SDK was at least 6x faster for read operations. Moremover, the Pinecone 3 SDK's query method does not use await like upsert so it cannot be optimized through async calls. Perhaps this is a limitation on the Pinecone DB query engine side where queries. But having batch support in Pinecone 3 would bring parity with the Pinecone 2 SDK and lead to much faster performance.
Describe alternatives you've considered
Potentially making query an asyc by adding an await, like with upsert. However there might be a limitation on the Pineceone side depending on how queries are treated that would not lead to optimizations
Who will this benefit?
Batch uses cases for the query method
Are you interested in contributing this feature?
Don't have the familiarity to contribute actual code, but can help with ideas
Is this your first time submitting a feature request?
Describe the feature
In the Pincone 2.x.x SDK, we have the ability to pass a batch of vectors to
index.querymethod. In the Pinecone 3 SDK, we can only pass a single vector toindex.query. We need high performance on the query method because we want to batch process a large vector dataset.It makes it very slow for reading operations. I found that the Pinecone 2 SDK was at least 6x faster for read operations. Moremover, the Pinecone 3 SDK's
querymethod does not useawaitlikeupsertso it cannot be optimized through async calls. Perhaps this is a limitation on the Pinecone DB query engine side where queries. But having batch support in Pinecone 3 would bring parity with the Pinecone 2 SDK and lead to much faster performance.Describe alternatives you've considered
Potentially making
queryan asyc by adding an await, like withupsert. However there might be a limitation on the Pineceone side depending on how queries are treated that would not lead to optimizationsWho will this benefit?
Batch uses cases for the query method
Are you interested in contributing this feature?
Don't have the familiarity to contribute actual code, but can help with ideas
Anything else?
No response