EmbeddingGemma 2: an open, lightweight multimodal embedding model
deepmind.google·6d ago
AI Engineering
Lobsters·Aug 26·5 min read
mold: A Massively Parallel Linker
TL;DR
Mold is a new Unix/Linux linker that dramatically speeds up the linking process for large C++ programs by applying systematic data parallelism. It achieves link times of under a second for multi-gigabyte binaries, outperforming existing linkers by up to 112 times.
✦ Why It Matters
Consider switching to mold for faster linking times in your C++ projects today.
Key Takeaways
- Evaluate mold for your next C++ project to reduce build times.
- Understand the benefits of data parallelism in linking.
- Compare mold's performance against lld and GNU ld.
- Explore the architectural constraints mold overcomes in linking.
How It Works
Mold's architecture separates symbol resolution from archive processing, allowing different parts of the linking process to run in parallel. This decoupling enables better utilization of CPU cores, significantly speeding up the overall linking time.