Fixing Ill-Formed UTF-16 Strings With SIMD, 9x faster
lemire · x · 2026-08-26
Daniel Lemire published a new paper on fixing ill-formed UTF-16 strings using SIMD instructions (ARM NEON and x64 SSE/AVX2/AVX-512). It achieves 18.9 GB/s on Apple M4 and 7.5 GB/s on Intel Ice Lake, using as little as 0.4 instructions per byte, up to nine times faster than the old method. The code is in Chromium browsers, and String.prototype.toWellFormed in Node.js 25 is about five times faster.
Related event: SIMD Repairs Malformed UTF-16 Strings 9x Faster(2 posts)→
More from Infra
- MiniCPM-V 4.7 runs a 35B MoE multimodal model on phones with just 3GB memory — nevrekaraishwa2 · 2026-10-15
- Nvidia Is Now Worth More Than the Entire Indian Stock Market — theteknosaur · 2026-10-15
- NVIDIA joins EU's AI in Science Enablers, offering compute, training and cloud resources — nordicinst · 2026-10-15
- Open source's hardware trap: data centers crowd out consumer GPUs, local LLMs become a privilege — Dante_77A · 2026-10-15
- NVIDIA backs EU AI in Science push as 35 supercomputers to serve 3M researchers — NVIDIA Blog · 2026-10-15
- Jensen Huang: 2027 AI bottleneck is 35GW of power, revenue justifies the capex — BG2 Pod · 2026-10-15