Top all-time · #3268
Casey Primozic
Personal blog · cprimozic.net/blog · on HN as Ameoprogrammingweb
Casey Primozic's personal blog at cprimozic.net covers programming and web. Since 2018, 23 different Hacker News readers have linked it 61 times, most often “Speeding Up the Webcola Graph Viz Library with Rust and WebAssembly”. It is still writing: the last post was on 2026-01-11, and you can follow it by feed.
Too few sampled comments (11) to say how often HN recommends it.?
Why HN links it
HN readers link Casey Primozic's blog for hands-on posts on neural networks, such as computing boolean logic and reverse-engineering a small network, plus posts on Geotoy's tech and a signal analyzer.
Summarised by Claude from a sample of 11 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
The homepage of Casey Primozic / Ameo
Most-linked posts?
- Speeding Up the Webcola Graph Viz Library with Rust and WebAssembly4 people · 5 mentions · first linked 2021-06-15
- What I've Learned Building Interactive Embedding Visualizations3 people · 4 mentions · first linked 2024-05-16
- Reverse engineering a neural network's clever solution to binary addition (2023)2 people · 5 mentions · first linked 2023-01-16
- Trying Kolmogorov-Arnold Networks in Practice2 people · 4 mentions · first linked 2024-07-02
- Exploring Neural Networks Visually in the Browser2 people · 2 mentions · first linked 2022-04-03
- Logic Through the Lens of Neural Networks2 people · 2 mentions · first linked 2022-08-14
- My favorite tools and techniques for procedural gamedev2 people · 2 mentions · first linked 2024-07-27
- Optimizing Advent of Code D9P2 with High-Performance Rust2 people · 2 mentions · first linked 2025-01-01
- Growing Bonsai Networks with RNNs2 people · 2 mentions · first linked 2023-08-07
- FM Synthesis in the Browser with Rust, Web Audio and WebAssembly with SIMD(2021)2 people · 2 mentions · first linked 2021-06-25
What Hacker News says?
That's something I'd use a canvas or even a SVG for. SVG FFT Analyzer sound like a fun project. Oh, looks like I'm way late: https://cprimozic.net/blog/building-a-signal-analyzer-with-m...
Comment linking /blog/building-a-signal-analyzer-with-modern-web-tech/
on “Why fastDOOM is fast”Submitted Fixing a bug in Google Chrome as a first-time contributor as a story · 645 points
HN discussionhttps://cprimozic.net/ Personal site and blog. Designed with the goal of being what I want people to see if they were to google me. My favorite part is my portfolio: https://cprimozic.net/portfolio/ It contains descriptions + screenshots of my significant projects going back to when I first learned programming.
Comment linking /portfolio/
on “Ask HN: Share your personal site”Submitted Self-Hosting Dozens of Web Applications and Services on a Single Server as a story · 511 points
HN discussion>Who will tell them how an LLM works and that the neural net does not calculate anything? You don't understand LLMs as well as you think you do. Yes, the neural network calculates things. >It only predicts the next token in a sentence of a calculation if it's been loss-minimized for that specific calculation. No that's not necessary at all. https://www.alignmentforum.org/posts/N6WM6hs7RQMKDhYjB/a-mec... https://cpri…
Comment linking /blog/reverse-engineering-a-small-neural-network/
on “Financial Statement Analysis with Large Language Models”Submitted My favorite tools and techniques for procedural gamedev as a story · 226 points
HN discussionI've tried out and written about[1] KANs on some small-scale modeling, comparing them to vanilla neural networks, as previously discussed here: https://news.ycombinator.com/item?id=40855028 . My main finding was that KANs are very tricky to train compared to NNs. It's usually possible to get per-parameter loss roughly on par with NNs, but it requires a lot of hyperparameter tuning and extra tricks in the KAN archite…
Comment linking /blog/trying-out-kans/
on “A new type of neural network is more interpretable”Submitted Speeding Up the Webcola Graph Viz Library with Rust and WebAssembly as a story · 218 points
HN discussion>People keep calling it "next next token predictors", but clearly there is something more going on and I would love for someone to give a simple explanation. Next token prediction is the objective function. The model is asked to predict the next word yes but it's also allowed to compute the answer and more importantly, the entire training process is supposed to be the model learning and figuring out what sort of com…
Comment linking /blog/reverse-engineering-a-small-neural-network/
on “How Chain-of-Thought Reasoning Helps Neural Networks Compute”Submitted My Evaluation of SvelteKit for Full-Stack Web App Development as a story · 194 points
HN discussion>that I can think of - is that the models "discovered" how to do arithmetic in a general way. I mean..yeah. Maybe i've simply misunderstood but It seems like you're presenting this as an outlandish idea but something like this is what Neural Networks regularly end up doing especially in the limit (as loss trends down to zero). https://cprimozic.net/blog/reverse-engineering-a-small-neura... https://www.alignmentforum…
Comment linking /blog/reverse-engineering-a-small-neural-network/
on ““Emergent” abilities in LLMs actually develop gradually and predictably – study”Submitted Trying Kolmogorov-Arnold Networks in Practice as a story · 146 points
HN discussion- nl · 2024-07-07
> LLMs by design are not capable of reason. This isn't true. A deep neural network certainly can emulate the logical functions we think of as "reasoning" (ie, AND/OR/XOR functions). See for example: https://cprimozic.net/blog/boolean-logic-with-neural-network... https://www.cs.toronto.edu/~axgao/cs486686_f21/lecture_notes... https://towardsdatascience.com/emulating-logical-gates-with-... https://medium.com/@stanleyd…
Comment linking /blog/boolean-logic-with-neural-networks/
on “Reasoning in Large Language Models: A Geometric Perspective” Submitted Habits I've developed for fast and efficient programming as a story · 108 points
HN discussionHow does Tinygrad fall short? Performance is fine [0]. It's much smaller than Pytorch and all, but that's kind of in the name. I've been hearing about MLIR and OpenXLA for years through Tensorflow, but I've never seen an actual application using them. What out there makes use of them? I'd originally hoped it'd allow Tensorflow to support alternate backends, but that doesn't seem to be the case. 0: https://cprimozic.…
Comment linking /notes/posts/machine-learning-benchmarks-on-the-7900-xtx/
on “ROCm Device Support Wishlist”Submitted Reverse engineering a neural network's clever solution to binary addition (2023) as a story · 87 points
HN discussionAMD's hardware might be compelling if it had good software support, but it doesn't. CUDA regularly breaks when I try to use Tensorflow on NVIDIA hardware already. Running a poorly-implemented clone of CUDA where even getting Pytorch running is a small miracle is going to be a hard sell. All AMD had to do was support open standards. They could have added OpenCL/SYCL/Vulkan Compute backends to Tensorflow and Pytorch a…
Comment linking /notes/posts/machine-learning-benchmarks-on-the-7900-xtx/
on “ROCm Device Support Wishlist”Submitted Exploring Neural Networks Visually in the Browser as a story · 84 points
HN discussionI found this article also interesting (I believe also from the author, but correct me if I'm wrong): https://cprimozic.net/blog/the-tech-behind-geotoy/#overview
Comment linking /blog/the-tech-behind-geotoy/
on “Geotoy – Shadertoy for 3D Geometry”Submitted Generating 3D Meshes from Text as a story · 33 points
HN discussion> Roughly speaking, the complexity of an object is the length of its shortest description. This has a ton of really interesting implications. Even though Kolmogorov complexity is technically incomputable due to the halting problem, you can estimate it. One method of doing that is to create a bunch of random Turing machines and see how often they produce some output string[1]. You have to cut them off after running f…
Comment linking /blog/boolean-logic-with-neural-networks/
on “Binary Lambda Calculus (2020)”Submitted Non-Pointless Software Projects for New Devs in the LLM Age as a story · 17 points
HN discussionhttps://cprimozic.net/blog/trying-out-kans/
Comment linking /blog/trying-out-kans/
on “Philosophical Thoughts on Kolmogorov-Arnold Networks (2024)”Submitted Logic Through the Lens of Neural Networks as a story · 11 points
HN discussionI did a little mini research project into trying out KANs in practice for some toy problems, also back in 2024 when they were the hot new thing: https://cprimozic.net/blog/trying-out-kans/ TL;DR KANs are tricker to train than traditional neural networks, but they largely have similar loss values given equivalent parameter counts. Part of this may be due to the fact that most of the optimizers and other components of…
Comment linking /blog/trying-out-kans/
on “Philosophical Thoughts on Kolmogorov-Arnold Networks (2024)”Submitted Reverse Engineering a Neural Network's Clever Solution to Binary Addition as a story · 7 points
HN discussionThis year, some members of the Rust Programming Language Community Server on Discord set out to solve AoC in under 1ms. I'm pleased to announce that through the use of LUTs, SIMD, more-than-questionable `unsafe`, assertions, LLVM intrinsics, and even some inline ASM that goal has been reached (for real this time)! *As of today, our solutions are able to solve all 49 problems in <1ms!* I have obtained consent from al…
Comment linking /blog/optimizing-advent-of-code-2024/
on “Solving Advent of Code 2024 in Under 1ms”Submitted Speeding Up Three.js with Depth-Based Fragment Culling as a story · 6 points
HN discussionI like the meme of the personal webpage, which has sort of fallen by the wayside with the rise of social media. It's one of those ideas that spreads because you see other people doing it: You get the OG blogs still up with 90s styling, student blogs with like 2 posts, and longform explorations of trying to squeeze out every ms of performance from D3 by rewriting parts of it in WASM [1], which link to the posts that…
Comment linking /blog/speeding-up-webcola-with-webassembly/
on “Nobody cares about your blog”Submitted Trying Out Cloudflare's `Foundations` Library for Rust as a story · 5 points
HN discussion
