I am using the GPU version of Milvus 2.4 standalone. I want to have the fastest possible query speed as that is what I need for my production use case.
I am using a few thousand records of dimension 1,024. I am using the GPU_IVF_FLAT index and my server is running slower than my laptop for these.
How do I speed it up? I am using a lot of defaults like default consistency. Does any of that matter?
Thanks.
I was hoping for faster query results and a big speed up with GPU vs CPU.
AINewbieNewYork is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.
- there is no reason to run ivf index on thousand entities.
- there is no reason to use GPU as well.
gpu is usually good at batch and large computation cases.
if you only have thousand vectors, graph or any kind of index could be fast to 1000+ QPS or even higher
james is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.