23 stories tagged simd, newest first.
34 points by StalwartLabs | 12 comments
Release announcement for Stalwart v1.0, an open-source Rust mail server, detailing SIMD/SWAR parser rewrites, storage tuning, a new search engine, and benchmarks against Postfix and Dovecot.
146 points by zdw | 47 comments
A blog post describing a SIMD-accelerated FAST corner detector written in assembly for the ESP32-S3, roughly doubling the reference implementation's performance.
118 points by martz | 40 comments
Blog post benchmarking VLC's Gaussian blur filter in C against a C# port via .NET Native AOT, across line-by-line, algorithmically optimized, and AVX2 SIMD versions.
243 points by pngmangi | 21 comments
Blog post examining the x86 FMA instruction vfmaddrnd231pd, which performs fused multiply-add with per-instruction rounding control.
47 points by zX41ZdbW | 5 comments
Blog post on purplesyringa explaining how to implement vectorized count-leading-zeros and count-trailing-zeros using floating-point tricks.
54 points by ravenical | 3 comments
Blog post by Sergey Davidoff on making SIMD abstractions in Rust safe internally, avoiding unsafe raw intrinsics.
188 points by pjmlp | 48 comments
Edward Kmett blog post on The Comonad.Reader titled Turbo Haskell, covering a Haskell compiler for the JVM: borrowing libraries, evaluation and concurrency, SIMD, and tail calls.
3 points by birdculture | 0 comments
Blog post by Sergey Davidoff surveying the state of SIMD libraries and support in Rust as of 2026, including Fearless SIMD, std::simd, wide, pulp, and macerator.
200 points by verdagon | 51 comments
Blog post by Sergey Davidoff surveying the state of SIMD libraries in Rust in 2026, including Fearless SIMD, std::simd, wide, pulp and macerator.
382 points by yurivish | 138 comments
Go Blog post describing experimental SIMD APIs in Go 1.26 and 1.27 for amd64, arm64 NEON, and wasm, and work toward platform-independent SIMD.
307 points by verdagon | 48 comments
Linebender blog post announcing the v1.0 release of fearless_simd, a Rust library for safe portable SIMD, and reflecting on its goals and design.
967 points by tosh | 702 comments
A post by Daniel Lemire on X discussing Xiaomi's Xring O3 CPU, its large cache, ARM C1-Ultra cores with SME2/SVE2, and performance compared to Apple chips.
568 points by WadeGrimridge | 211 comments
Mitchell Hashimoto blog post arguing SIMD is simple to learn and showing the common shape of SIMD code for speeding up loops.
379 points by tombh | 153 comments
All The Views presents the longest lines of sight on Earth found by an exhaustive algorithm, topped by a 530km view from the Hindu Kush to Pik Dankova.
396 points by flykespice | 133 comments
GitHub repository from the FFmpeg project containing lessons teaching assembly language programming as used in FFmpeg, with translations.
341 points by harambae | 96 comments
Tom's Hardware article reporting FFmpeg developers achieving a roughly 100x speedup in a function via handwritten assembly code.
774 points by davikr | 206 comments
First lesson in FFmpeg's GitHub-hosted course teaching assembly language programming as used in the FFmpeg multimedia project.
926 points by k__ | 689 comments
Simon Willison blog post on a llama.cpp PR doubling WASM speed via SIMD, mostly written by DeepSeek R1, plus his own results using R1 for code.
585 points by atombender | 196 comments
A long blog post by Ragnar Groot Koerkamp implementing and optimizing a static S+ search tree for high-throughput sorted-data search, reaching up to 40x faster than binary search.
372 points by p1esk | 69 comments
Blog tutorial implementing high-performance CPU matrix multiplication in about 150 lines of C, following the BLIS design and outperforming NumPy/OpenBLAS on a Ryzen 7700.
368 points by weinzierl | 72 comments
GitHub pull request to llama.cpp by jart that upstreams llamafile's vectorized expf() functions to compute SiLU and softmax faster and more accurately on CPU.
1286 points by lawrencechen | 413 comments
Justine Tunney's blog post describing 84 new matrix multiplication kernels that make llamafile prompt evaluation 30-500% faster on CPUs than llama.cpp.
406 points by ingve | 70 comments
Blog post by Miguel Young de la Sota explaining how to design a fast, portable SIMD algorithm in Rust, using base64 decoding as the example.