-
Notifications
You must be signed in to change notification settings - Fork 2.2k
docs: indexers benchmark #4003
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
docs: indexers benchmark #4003
Conversation
Latency summaryCurrent PR yields:
Breakdown
Backed by latency-tracking. Further commits will update this comment. |
let me restructure the TOC a bit |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
When you talk about 1k query vectors it is not clear if this is 'done as a batch' (one function call that processes 1k query vector) or if query vectors are processes one at a time. If I were a user probably I would like to know
- What's the expected time between
.search
and getting a result for a single query? - Whats the expected throughput of queries I can expect using
X
resources indelta
time.
The document seems to answer the first question since it states Then we search with the respective search set, using a batch size of `1`, to mimic single query operations.
The second question is not as relevant but still interesting for users with really high traffic.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I like it so far. Obviously the results data is needed.
In the final version, I'd put the results at the top and put the methodology behind. Most views will be for the results and thus they should be at the top and easiest accessible.
@maximilianwerk moved results to top. @hanxiao @davidbp added results now too. Can you check again? |
Not sure why the deployment doesn't include the subpage. Locally it is there |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I see now the benchmarks.
In the netlify deployment? I still don't. Tried a hard refresh but nothing. |
me neither. weird. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I find it confusing, that the x-axis has varying scales in all the plots. This makes it much harder to get fast conclusions. Having a changing y-axis is OK. Opinions?
I'd need to recreate them and the code from @Hippopotamus0308 is not yet in the benchmarks repo. We can either wait and recreate them with fixed axes, or merge it as it is now. |
@cristianmtr I've added to the pr in https://github.com/jina-ai/benchmark_indexers/pull/33, please check it. |
I tried to render them with fixed x_range but they are not all readable ex. |
Ok then don't do that and keep it as they are :) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
apart from this and the precision/recall, I am good with merging.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
📝 Docs are deployed on https://docs-indexers-benchmark--jina-docs.netlify.app 🎉 |
TODO
iframe
? Can we include html files somehow?