EmbeddingGemma 2: an open, lightweight multimodal embedding model
deepmind.google·6d ago
AI Engineering
Lobsters·Jul 24·5 min read
ffi_call_plan caching for GLib
TL;DR
Anthony Green introduced a caching mechanism called 'call plan' to libffi, significantly reducing overhead in GClosure invocations by up to 25%. This enhancement streamlines function calls and improves performance in subsequent executions.
✦ Why It Matters
Update your GLib build to include the latest libffi for improved performance.
Key Takeaways
- Implement call plan caching in your projects for better performance.
- Monitor GClosure invocation overhead to assess improvements.
- Utilize frame-pointers for effective profiling across FFI boundaries.
- Stay updated on GLib merge requests for new features.
How It Works
The 'call plan' mechanism caches the results of function call preparations, allowing subsequent calls to skip redundant computations. This is particularly beneficial for GClosure invocations, which often involve complex setup processes.
By storing the necessary information from the first call, the system can significantly reduce the time and resources required for future invocations.