Load RAG database in memory on use. #1344
Open
+60
−2
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This change uses a trick to force the RAG database into memory by
dumping the SQLite database from file to a temporary in-memory version
of it using SQLite's `backup` function.
Based on `percall` statistic, this almost halves execution times.
With on-disk database.
```
ncalls tottime percall cumtime percall filename:lineno(function)
911 0.017 0.000 47.983 0.053 .../src/codegate/storage/storage_engine.py:149(search)
911 41.336 0.045 41.336 0.045 {method 'execute' of 'sqlite3.Cursor' objects}
911 0.007 0.000 6.489 0.007 .../src/codegate/inference/inference_engine.py:90(embed)
```
With in-memory database.
```
ncalls tottime percall cumtime percall filename:lineno(function)
658 0.009 0.000 21.198 0.032 .../src/codegate/storage/storage_engine.py:147(search)
658 16.458 0.025 16.458 0.025 {method 'execute' of 'sqlite3.Cursor' objects}
658 0.004 0.000 4.638 0.007 .../src/codegate/inference/inference_engine.py:90(embed)
```
Also, added profiling to providers and search. Profiling can be
activated exporting `CODEGATE_PROFILE_<key>` where `<key>` is the
string passed to `@profiled` annotations. Profiling points are
statically defined throughout the codebase.
blkt
force-pushed
the
perf/in-memory-rag
branch
from
2ed45bd to
3b3b945
Compare
|
Nice, taking a look at this, what is the memory footprint like? |
|
I'm sorry @lukehinds I missed your comment. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This change uses a trick to force the RAG database into memory by dumping the SQLite database from file to a temporary in-memory version of it using SQLite's
backupfunction.Based on
percallstatistic, this almost halves execution times.With on-disk database.
With in-memory database.
Also, added profiling to providers and search. Profiling can be activated exporting
CODEGATE_PROFILE_<key>where<key>is the string passed to@profiledannotations. Profiling points are statically defined throughout the codebase.