harshakhmk

But, I needed to store vectors on disk, as I need to keep RAM consumption as minimum as possible, is there a way for it Even in the client code,...

> If you don't use payloads `on_disk_payload` is unrelated. Instead you want to set `on_disk: true` on vectors. > > And even then, it will still utilize all available RAM...

``` let dimension= 785 let rows=20000 let mut collection_name="col" let config= QdrantClientConfig::from_url("localhost") let client= QdrantClient::new(Some(config)).unwrap() let x= client.create_collection(& CreateCollection{ collection_name: collection_name, vectors_config: Some(VectorsConfig{ config:Some(Confi::Params(VectorParams{ size: dimension, distance:cosine, on_disk:Some(true), .. Default::...

Let me know, if the above sample code is missing optimisation or any other thing is missing regarding optimisation, I have shared the base code

> > basically my device has low RAM capacity, I needed to keep vectors on disk, I want to reduce the unused RAM > > If that is the only...

> @harshakhmk how do you measure memory consumption? If you are using htop or similar tool, it is likely that you are using memory + disk cache I have checked...

I have observed for a couple of time, still memory space is not reduced

I tried the above command ``` cat /proc//status | grep Rss ``` But didn't notice any change in performance in terms of RAM & Cpu

@generall @timvisee , any leads on how to get the results of our optimisation would be highly appreciated I would also like to know what tools you suggest to track...

I observation, I have noticed is when I configure payload to on_disk=true for indexing/quantization/storing vectors, the RAM is increasing which shouldn't be the case, as all data will be on...