Initial cut for a cuVS Java API #450
Conversation
Co-authored-by: Vivek Narang <vivek@searchscale.com>
|
FYI @cjnolet ^ |
cjnolet
added
improvement
|
/ok to test |
narangvivek10
force-pushed
the
java-api
branch
from
2cfab29 to
1597158
Compare
chatman
changed the title
|
@naramgvivek10 Let's move CuVSResources to the cuvs package instead of common? That way we can abstract out the internals of Panama out of sight of the users. |
…ent for data array
| @@ -1,5 +1,6 @@ | |||
| ## common | |||
| __pycache__ | |||
| .gitignore | |||
There was a problem hiding this comment.
Please don't check .gitignore into .gitignore. We need to know when this file changes.
| @@ -0,0 +1,7 @@ | |||
| export CMAKE_PREFIX_PATH=`pwd`/../cpp/build | |||
There was a problem hiding this comment.
Can we put this logic in the top-level build.sh please? We do that with all of the other language support.
| @@ -0,0 +1,697 @@ | |||
| /* | |||
There was a problem hiding this comment.
I know we've discussed having this automatically generate on each build in a future iteraiton. Can you please create a cuVS Github issue to do that?
| public void testIndexingAndSearchingFlow() throws Throwable { | ||
|
|
||
| // Sample data and query | ||
| float[][] dataset = {{ 0.74021935f, 0.9209938f }, { 0.03902049f, 0.9689629f }, { 0.92514056f, 0.4463501f }, { 0.6673192f, 0.10993068f }}; |
There was a problem hiding this comment.
Hardcoding tests makes it very hard to debug and troubleshoot problems, and they also become very brittle because we've found changes in the compute architectures can introduce flakiness and downright failures from hardcoded tests. Please generate the data instead of hardcoding the values and use bfknn to validate.
There was a problem hiding this comment.
We added a randomized test in my latest commit. We noticed for topK > 64 the returned values are incorrect (neighbor values are in the range outside of the dataset size).
… etc. (#6) * Ability to configure CAGRA compression parameters * Enabling indexing threads * Enabling RMM pool resource configuration * Bug fixes for wrong values passed in index and search parameters * Deallocation of resources using Autoclosable * Including a randomized test Co-authored-by: Vivek Narang <vivek@searchscale.com> Co-authored-by: Ishan Chattopadhyaya <ishan@apache.org> Co-authored-by: Puneet Ahuja <puneet@searchscale.com>
Co-authored-by: Vivek Narang <vivek@searchscale.com>
…eparate example project
…ent for data array
… etc. (#6) * Ability to configure CAGRA compression parameters * Enabling indexing threads * Enabling RMM pool resource configuration * Bug fixes for wrong values passed in index and search parameters * Deallocation of resources using Autoclosable * Including a randomized test Co-authored-by: Vivek Narang <vivek@searchscale.com> Co-authored-by: Ishan Chattopadhyaya <ishan@apache.org> Co-authored-by: Puneet Ahuja <puneet@searchscale.com>
* Bruteforce API implementation Co-authored-by: Vivek Narang <vivek@searchscale.com>
A Java API for cuVS for easy integration into Apache Lucene or other Java based projects.
Try:
For generating docs,
mvn javadoc:javadocPrerequisites:
Todo: