Top all-time · #3049
Ash's Blog
Personal blog · ashvardanian.com · on HN as ashvardanianprogrammingai
Ash's Blog is a personal blog at ashvardanian.com about programming and ai. Since 2023, 25 different Hacker News readers have linked it 75 times, most often “Show HN: I've Published 28B Molecule Embeddings on AWS Open Data”. It is still writing: the last post was on 2026-10-21, and you can follow it by feed.
Too few sampled comments (5) to say how often HN recommends it.?
Why HN links it
HN readers link Ash's posts on SIMD, vector instructions, and fast C/C++ and Python similarity search and benchmarking. One commenter calls his molecule-search write-up a great read.
Summarised by Claude from a sample of 5 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
CS/AI Researcher building high-performance infrastructure. Founder of Unum Cloud. Writing about Tech, Startups, and making software 'Less Slow'.
Most-linked posts?
- Show HN: I've Published 28B Molecule Embeddings on AWS Open Data4 people · 4 mentions · first linked 2023-11-21
- Processing Strings 109x Faster Than Nvidia on H1004 people · 4 mentions · first linked 2025-09-17
- Before AI's Kepler Moment3 people · 5 mentions · first linked 2025-10-02
- Beyond OpenMP in C++ and Rust: Taskflow, Rayon, Fork Union3 people · 4 mentions · first linked 2025-05-22
- Binding a C++ Library to 10 Programming Languages3 people · 3 mentions · first linked 2023-11-10
- 2x Faster Hashes on AWS Graviton: Neon → SVE23 people · 3 mentions · first linked 2025-10-07
- NumPy vs. BLAS: Losing 90% of Throughput3 people · 3 mentions · first linked 2024-03-13
- Show HN: SimSIMD vs. SciPy: How AVX-512 and SVE make SIMD cleaner and ML faster2 people · 5 mentions · first linked 2023-10-07
- JavaScript for AI: Array or TypedArray for Vector Search2 people · 4 mentions · first linked 2023-10-23
- Show HN: Python, C, Assembly – 2'500x Faster Cosine Similarity2 people · 3 mentions · first linked 2023-10-30
What Hacker News says?
Hey! First of all, cool to see some activity on this front! I’ve written a fair share of pure CPython bindings and regularly post about implementing them with minimal overhead (< https://ashvardanian.com/posts/discount-on-keyword-arguments... >) and would love to share a few recommendations, questions, and concerns :) Just a suggestion to help you grow—I'd restructure the landing page (< https://hpyproject.org/ >) a…
Comment linking /posts/discount-on-keyword-arguments-in-python/
on “HPy – A better C API for Python”Submitted Full Unicode Search at 50× ICU Speed with AVX‑512 as a story · 216 points
HN discussionThere are several ways to implement it in NumPy, often resulting in 20% variance. I've added a reference implementation to my mirror of the blogpost and the Modular team will soon update the original posting as well: https://ashvardanian.com/posts/understanding-simd-complexity...
Comment linking /posts/understanding-simd-complexity/
on “Understanding SIMD: Infinite complexity of trivial problems”Submitted Processing Strings 109x Faster Than Nvidia on H100 as a story · 216 points
HN discussionThree separate passes over JavaScript arrays are quite costly, especially for high-dimensional vectors. I'd recommend using `TypedArray` with vanilla `for` loops. It will make things faster, and will allow using C extensions, if you want to benefit from modern hardware features, while still implementing the logic in JavaScript: https://ashvardanian.com/posts/javascript-ai-vector-search/
Comment linking /posts/javascript-ai-vector-search/
on “How does cosine similarity work?”Submitted Beyond OpenMP in C++ and Rust: Taskflow, Rayon, Fork Union as a story · 127 points
HN discussionYou are welcome :) There are more materials in my blog, including 4 articles from October 2023: https://ashvardanian.com/archives/ Benchmarks are quite easy to reproduce, especially in Python. The numbers for AWS Graviton 3, Apple M2 and Intel Sapphire Rapids CPUs can be found here: https://ashvardanian.com/posts/simsimd-faster-scipy/
Comment linking /posts/simsimd-faster-scipy/
on “SimSIMD: Hardware-accelerated similarity metrics and distance functions”Submitted Processing Strings 109x Faster Than Nvidia on H100 as a story · 34 points
HN discussionYes, the improvements for float32 aren't very dramatic, but it can be 3x NumPy/SciPy: https://ashvardanian.com/posts/simsimd-faster-scipy/ C11 is also a bit tricky, as its support is optional, as far as I remember.
Comment linking /posts/simsimd-faster-scipy/
on “SimSIMD: Hardware-accelerated similarity metrics and distance functions”Submitted Abusing Vector Search for Texts, Maps, and Chess ♟ as a story · 18 points
HN discussionEasy. Here is an article I wrote in 2016: https://ashvardanian.com/posts/whats-wrong-with-wwdc-2016/
Comment linking /posts/whats-wrong-with-wwdc-2016/
on “Thoughts on the New iOS Control Centre”Submitted Stringzilla v4 Introduces 500 GigaCUPS Edit Distance on GPUs as a story · 12 points
HN discussionIt’s a nice post, but “using array methods” probably shouldn’t be placed in the “Efficient Implementation” section. As often happens with high-level languages, a single plain old loop is faster than three array methods. Similarly, if you plan to query those vectors in search, you should consider continuous `TypedArray` types and smaller scalars than the double precision `number`. I know very little about JS, but som…
Comment linking /posts/javascript-ai-vector-search/
on “How to Implement a Cosine Similarity Function in TypeScript”Submitted Show HN: I've Published 28B Molecule Embeddings on AWS Open Data as a story · 10 points
HN discussionTo accelerate the analytical solution, there are more obvious techniques, like replacing trigonometric opetations with series. This is a 10x improvement in https://ashvardanian.com/posts/google-benchmark/#slow-trigon...
Comment linking /posts/google-benchmark/
on “When Greedy Algorithms Can Be Faster [C++]”Submitted From Dating to Vector Search – “Stable Marriages” on a Global Scale as a story · 9 points
HN discussionARMs Neoverse IP does support SVE, so it's at least already relevant in cloud applications. Apparently AWS Graviton3 has 256bit vectors, but Graviton4 regressed back to 128bit vectors for some reason? https://ashvardanian.com/posts/aws-graviton-checksums-on-neo...
Comment linking /posts/aws-graviton-checksums-on-neon-vs-sve/
on “AVX2 is slower than SSE2-4.x under Windows ARM emulation”Submitted Scaling Elections with GPUs and Mojo as a story · 8 points
HN discussionNice project! A regular on HN and creator of usearch built an embedding search for the same dataset and did a write up which is a great read. https://ashvardanian.com/posts/usearch-molecules/
Comment linking /posts/usearch-molecules/
on “Show HN: I wrote a GPU-less billion-vector DB for molecule search (live demo)”Submitted The Longest Nvidia PTX Instruction as a story · 8 points
HN discussionThree fundamental flaws of SIMD ISA:s - Bits'n'Bites ( https://news.ycombinator.com/item?id=43783416 ) 1. Snell-Pym » Processor architecture ( https://www.snell-pym.org.uk/archives/2017/03/17/processor-a... ) 2. Fun with vectors in the Raspberry Pi 1 - Part 9 ( https://thinkingeek.com/2021/08/22/raspberry-vectors-part-9/ ) I think we have enough pieces of machinery working already that we can start with the most exc…
Comment linking /posts/understanding-simd-complexity/
on “Show HN: Discover and link the indieweb with a simple text query”Submitted Fork Union: Beyond OpenMP in C++ and Rust? as a story · 7 points
HN discussionIf the author of the article sees this, this link in the table of contents doesn't work: https://ashvardanian.com/posts/python-c-assembly-comparison/...
Comment linking /posts/python-c-assembly-comparison/
on “Python, C, Assembly – Faster Cosine Similarity”Submitted Fork Union: Beyond OpenMP in C++ and Rust? as a story · 7 points
HN discussionHaven’t seen that, thanks for sharing! I was thinking about implementing a set-search and clustering of similar nature here: https://ashvardanian.com/posts/abusing-vector-search/
Comment linking /posts/abusing-vector-search/
on “Stargazers intersections for most popular GitHub projects in Venn diagrams”Submitted What's Wrong with C++ Strings? as a story · 7 points
HN discussionhttps://ashvardanian.com
Comment linking /
on “Ask HN: Could you share your personal blog here?”Submitted Show HN: Python, C, Assembly – 2'500x Faster Cosine Similarity as a story · 7 points
HN discussionAlways pleasure to see short implementations, but I’d still take HNSW over such approach. Its core implementation can also be quite compact. In our case with USearch [1] its about 2k LOC, even with a custom implementation of priority queues and custom numeric types like uint40_t for indexes beyond 4B points. Obviously, together with bindings for Rust, Python, JavaScript, Java, Objective-C, Swift, and Wolfram it gets…
Comment linking /posts/abusing-vector-search
on “FANN: Vector Search in 200 Lines of Rust”Submitted Processing Strings 109x Faster Than Nvidia on H100 as a story · 6 points
HN discussion
