Tag
A technical blog post showing how to transpose vec3/vec4-style SIMD code into lane-based layouts so math operations scale beyond 4-wide XMM registers to 8- or 16-wide AVX/AVX-512 vectors, demonstrated with dot products and a CPU frustum culler on Zen 3.
Modular 社区会议录像已上线 YouTube，社区成员 Brian 展示了他的 Community Grant 项目 EmberJSON——一个受 Rust serde 启发的 Mojo 序列化框架，以及基于编译期反射和 SIMDJSON 风格索引的快速 JSON 解析器，并预告了关于新 Max 许可证的直播 Q&A。
Stalwart v1.0 is the first stable release of a Rust-based mail and collaboration server (JMAP, IMAP, POP3, SMTP, CalDAV, CardDAV, WebDAV) after five years of development, focusing on major performance gains via SIMD/SWAR optimizations and a stable database schema.
作者为 2 美元的 ESP32-S3 微控制器编写了利用 128 位 SIMD 指令加速的 FAST 特征检测器，性能约为参考实现的两倍，处理 QVGA 帧约需 6ms，并顺带开发了自用的寄存器分配器 basm。
ZXC 0.15.0 正式发布，这是一款面向“一次压缩、多次解压”场景的无损压缩库 libzxc，采用 C 编写，解压速度可达每秒数 GB，其各级别的解码速度均为 LZ4 的 1.5–2.8 倍，且输出体积相等或更小。
Timescale 开源了 pg_vectorsearch，一个 PostgreSQL 扩展，其 PRISM 索引在查询和建索引速度上均优于 pgvector 的 IVFFlat 和 HNSW，支持 RaBitQ 量化、SIMD 优化和动态更新。
Niles Salter’s talk introduces how to use SIMD data parallelism and bit manipulation primitives (CTZ, TZCNT, etc.) to accelerate a Zig tokenizer, achieving a throughput of 820 MB/s, and explores the potential for further speedups with AVX-512.
Sergey 'Shnatsel' Davidoff explains how Rust's SIMD abstractions can now be made safe on the inside, leveraging Rust 1.87's target_feature tracking and CPU feature tokens to eliminate pervasive unsafe blocks around SIMD intrinsics while keeping a+b ergonomics.
Anton Zhiyanov (@ohmypy) 报道 SQLite 核心团队正在开发 Vec1——SQLite 原生向量搜索（ANN）功能，支持欧氏（L2）与余弦距离，并使用 SIMD 加速向量运算；该功能已基本完成，仅剩测试改进等工作。
一篇技术博客，介绍如何利用浮点数指数相当于对数这一特性，用 AVX2 浮点指令实现可向量化的 clz（前导零计数）和 ctz（尾随零计数）多播（polyfill），并在 Haswell/Alder Lake 上给出优于标量版本的性能数据。
Hillock 0.8 is an open-source tool that replaces vector databases for local RAG using SQLite relational facts and 10,000-dimensional bit-packed hypervectors with SIMD/popcount gating, blocking hallucinated answers before token generation under 1.2GB VRAM.
Halide introduces wpd, a Rust-based WebP decoder that is memory-safer and significantly faster than libwebp, with up to 3.19x multi-threaded performance gains and near-full feature parity.
The article surveys the state of SIMD support in Rust in 2026, discussing advancements in SIMD libraries and implementation details.
Go introduces experimental SIMD APIs to enable platform-independent SIMD operations, aiming to provide near-asm performance across different CPU architectures.
Fearless SIMD v1.0 is released, offering safe and high-performance SIMD abstractions in Rust that minimize unsafe code while providing portable operations and platform intrinsics.
The author developed a faster Rust crate, sha1dc, for SHA-1 collision detection that runs at 68–81% of plain SHA-1 speed, which can make git pack verification twice as fast compared to existing solutions.
The article describes how using the SVE2 SIMD instruction on ARM processors can optimize JSON parsing in the simdjson library by leveraging a character matching instruction for faster performance.
Google shares open-source vectorized Quicksort code that achieves 10x speedup over C++ std::sort by leveraging SIMD instructions, offering performance-portable sorting across modern CPU architectures.
The article describes optimizations to the `gearhash` Rust crate, adding a NEON backend that doubles performance on ARM64 architectures, improving content-defined chunking for applications like Hugging Face's Xet client.
Singeli is a domain-specific language for high-performance programming with SIMD support, implemented in BQN and emitting C code, designed for flexible abstractions over CPU instructions.