EmbeddingGemma 2: an open, lightweight multimodal embedding model
deepmind.google·6d ago
AI Engineering
Lobsters·22d ago·5 min read
The last mile of a long road: faster NumPy in the browser
TL;DR
Emscripten-forge has enabled accelerated NumPy in the browser by linking OpenBLAS in WebAssembly, significantly improving matrix multiplication performance. This advancement allows for more efficient scientific computing directly in web environments.
✦ Why It Matters
Start using Emscripten-forge to enhance your web applications with accelerated scientific computing capabilities today.
Key Takeaways
- Explore Emscripten-forge for accelerated NumPy in your web projects.
- Utilize OpenBLAS for improved matrix operations in the browser.
- Consider the upcoming OpenBLAS features for enhanced performance.
- Leverage the language-agnostic capabilities of Emscripten-forge.
Full Summary
Running NumPy in the browser traditionally lacked performance optimizations due to the absence of an accelerated Basic Linear Algebra Subprograms (BLAS) library, resulting in slower matrix multiplications. Emscripten-forge has now integrated OpenBLAS with WebAssembly, allowing for substantial performance gains, particularly at matrix sizes like n = 1024.
The upcoming OpenBLAS release will further enhance performance with contributions from QuantStack and an optional Relaxed SIMD build for compatible engines. Emscripten-forge serves as a language-agnostic software distribution, expanding beyond Python and R to include C++, OCaml, and Lua, among others.
This builds on the foundational work of Pyodide and WebR, creating a more versatile ecosystem for scientific computing in the browser. The integration of a package manager also simplifies the installation and management of these tools.