4 ms·

I’ve done something similar (heh) in pure Postgres: https://github.com/turbo/pg-costop https://github.com/turbo/pg-costop

jononor 5y ago
Nice, useful specially on managed services where installing additional extensions may not be an option. If you want wider usage, you may want to add a section on how to install/setup this with a typical postgresql setup? And also, how was the code tested for correctness, and what is the expected performance?
phenkdo 5y ago
how does this do with large-scale searches >10M+ rows? any benchmarks on performance?
- minxomat 5y ago
  Oh it's terrible. More educational. You'd never ever want to do a full-scan cosine matching in production. Use locality sensitive hashing (with optimizations like amplification, SuperBit or DenseFly) for real world workloads.
- gk1 5y ago
  Since the answer, in minxomat's words, is "terrible," maybe look at Pinecone (https://www.pinecone.io https://www.pinecone.io) which makes light work of searching through 10M+ rows. It sits alongside, not inside, your Postgres (or whatever) warehouse. Disclosure: I work there.
  - fzysingularity 5y ago
    Out of curiosity, what kind of performance do you get with 100M rows with pinecone? Looking at your pricing tiers, ~100M rows would need ~200GB memory, and @ $0.1 / GB / hr that's $20 / hr if I'm not mistaken? Also, can you join with existing SQL tables to do hybrid searches in pinecone?
    - gk1 5y ago
      Our performance is independent of the collection size, thanks to dynamic sharding. You can expect 50-100ms pretty much regardless of size. We don't support external SQL joins yet. Depending on what you're trying to do, we have an upcoming feature that might do the job. At 100M vectors you're well into "volume discount" territory. Even more so with 3B vectors, as you mentioned in another comment. The free trial is capped at 10GB but shoot me an email (greg@pinecone.io) to get around it for a proper test and pricing estimate.
- fzysingularity 5y ago
  Others have pointed this out already, but have a look at Milvus (https://github.com/milvus-io/milvus https://github.com/milvus-io/milvus). I was able to get a simple version of it running with searches over ~3B vectors in under a second on a single machine (with out of the box configuration, and practically no optimization done just yet).

jononor 5y ago

Nice, useful specially on managed services where installing additional extensions may not be an option. If you want wider usage, you may want to add a section on how to install/setup this with a typical postgresql setup? And also, how was the code tested for correctness, and what is the expected performance?

phenkdo 5y ago

how does this do with large-scale searches >10M+ rows? any benchmarks on performance?

minxomat 5y ago
Oh it's terrible. More educational. You'd never ever want to do a full-scan cosine matching in production. Use locality sensitive hashing (with optimizations like amplification, SuperBit or DenseFly) for real world workloads.
gk1 5y ago
Since the answer, in minxomat's words, is "terrible," maybe look at Pinecone (https://www.pinecone.io https://www.pinecone.io) which makes light work of searching through 10M+ rows. It sits alongside, not inside, your Postgres (or whatever) warehouse. Disclosure: I work there.
- fzysingularity 5y ago
  Out of curiosity, what kind of performance do you get with 100M rows with pinecone? Looking at your pricing tiers, ~100M rows would need ~200GB memory, and @ $0.1 / GB / hr that's $20 / hr if I'm not mistaken? Also, can you join with existing SQL tables to do hybrid searches in pinecone?
  - gk1 5y ago
    Our performance is independent of the collection size, thanks to dynamic sharding. You can expect 50-100ms pretty much regardless of size. We don't support external SQL joins yet. Depending on what you're trying to do, we have an upcoming feature that might do the job. At 100M vectors you're well into "volume discount" territory. Even more so with 3B vectors, as you mentioned in another comment. The free trial is capped at 10GB but shoot me an email (greg@pinecone.io) to get around it for a proper test and pricing estimate.
fzysingularity 5y ago
Others have pointed this out already, but have a look at Milvus (https://github.com/milvus-io/milvus https://github.com/milvus-io/milvus). I was able to get a simple version of it running with searches over ~3B vectors in under a second on a single machine (with out of the box configuration, and practically no optimization done just yet).