Ultra Edge News

We still can't stop plagiarism in undergraduate computer science (2018) 14 by wonger_ | 22 comments on Hacker News.

For Some Recent Graduates, the A.I. Job Apocalypse May Already Be Here 5 by demirbey05 | 1 comments on Hacker News.

We're beating $359M in funding with two people and OCaml 18 by imadj | 3 comments on Hacker News.

Irritation Is Inspiration 14 by j4mehta | 6 comments on Hacker News.

Catbench Vector Search Demo Has Postgres SQL Throughput, Latency Monitoring Now 3 by tanelpoder | 0 comments on Hacker News.

Web Bench: a new way to compare AI browser agents 13 by suchintan | 3 comments on Hacker News.

Airlines are charging solo passengers higher fares than groups 85 by _tqr3 | 82 comments on Hacker News.

A toy RTOS inside Super Mario Bros. using emulator save states 10 by notorious_pgb | 3 comments on Hacker News. This started as a throwaway metaphor in a blog post, but is now fully runnable: a toy RTOS with preemptive multitasking inside of Super Mario Bros. on the NES. Essentially, this is: - A rudimentary preemptive RTOS - Using an unmodified NES emulator (FCEUX) as the CPU - "Unmodified" depending on how you define terms - With emulator save states as the thread contexts - With support for (very basic) mutexes, interrupt masking, and condition variables - Demonstrated using Super Mario Bros. 1-1 with sections of the map dedicated to various synchronization primitives There are many simplifications and shortcuts taken (doesn't even have task priorities), and it doesn't map 1:1 to true multithreading (e.g., emulator save states represent the state of the entire machine including RAM, whereas thread contexts represent a much more minimal slice), but I think it...

Why is it so hard to get families to live in community houses? 32 by caser | 35 comments on Hacker News.

Show HN: My LLM CLI tool can run tools now, from Python code or plugins 38 by simonw | 19 comments on Hacker News.

Why the Original Macintosh Had a Screen Resolution of 512×324 23 by ingve | 4 comments on Hacker News.

Show HN: Maestro – A Framework to Orchestrate and Ground Competing AI Models 3 by defqon1 | 0 comments on Hacker News. ive spent the past few months designing a framework for orchestrating multiple large language models in parallel — not to choose the “best,” but to let them argue, mix their outputs, and preserve dissent structurally. It’s called Maestro heres the whitepaper https://ift.tt/jfazy12 (Narrative version here: https://ift.tt/UDs5ck3... ) Core ideas: Prompts are dispatched to multiple LLMs (e.g., GPT-4, Claude, open-source models) The system compares their outputs and synthesizes them It never resolves into a single voice — it ends with a 66% rule: 2 votes for a primary output, 1 dissent preserved Human critics and analog verifiers can be triggered for physical-world confirmation (when claims demand grounding) The feedback loop learns not only from right/wrong outputs, but from what kind of disagreements lead to deeper truth Maestro isn’t a product or API — it’s a proposa...

Mutmut – Python Mutation Tester 3 by e-topy | 0 comments on Hacker News.

Advice to Tenstorrent 10 by lexoj | 6 comments on Hacker News.

Ask HN: What are you working on? (May 2025) 19 by david927 | 50 comments on Hacker News. What are you working on? Any new ideas that you're thinking about?

Gemma 3n Architectural Innovations – Speculation and poking around in the model 5 by nolist_policy | 0 comments on Hacker News.

Use ramoops for logging under Linux (2021) 3 by zdw | 0 comments on Hacker News.

Machinist and Machine 3 by throwaway71271 | 1 comments on Hacker News.

Trump administration halts Harvard's ability to enroll international students 250 by S0y | 172 comments on Hacker News.

We’ll be ending web hosting for your apps on Glitch 22 by js4ever | 4 comments on Hacker News.

An upgraded dev experience in Google AI Studio 15 by meetpateltech | 2 comments on Hacker News.

Gemma 3n preview: Mobile-first AI 71 by meetpateltech | 48 comments on Hacker News.

Red Programming Language 10 by hotpocket777 | 3 comments on Hacker News.

WireGuard-vanity-keygen: WireGuard vanity key generator 9 by simonpure | 1 comments on Hacker News.

Show HN: A native Hacker News reader with integrated todo/done tracking 5 by coolwulf | 2 comments on Hacker News. Hey HN! I'm excited to share a tool I've been working on - a native Hacker News reader built with Rust and egui. Here's a screenshot: https://ift.tt/Qe0BhaC... . As a daily HN reader, I've always struggled with keeping track of interesting posts I want to read later. Browser tabs pile up, bookmarks get forgotten, and I lose track of what I've already read. I needed a way to: 1. Browse HN efficiently (across all sections - hot, new, show, ask, jobs, best) 2. Quickly mark posts as "todo" for later reading 3. Mark posts as "done" when finished 4. Filter and search effectively I couldn't find a tool that combined all these features, so I built one. It's been tremendously helpful for my own HN reading workflow, and I thought others might find it useful too. Features: - *Integrated todo tracking*: Mark stories as "todo...

K-Scale Labs: Open-source humanoid robots, built for developers 7 by rbanffy | 1 comments on Hacker News.

The Fall of Roam 26 by ingve | 1 comments on Hacker News.

A Simulation in C++ of Joseph Weizenbaum's 1966 Eliza 3 by m1guelpf | 0 comments on Hacker News.

ClojureScript forks Google Closure to guarantee backward compatibility 33 by Borkdude | 3 comments on Hacker News.

Show HN: KVSplit – Run 2-3× longer contexts on Apple Silicon 66 by dipampaul17 | 7 comments on Hacker News. I discovered that in LLM inference, keys and values in the KV cache have very different quantization sensitivities. Keys need higher precision than values to maintain quality. I patched llama.cpp to enable different bit-widths for keys vs. values on Apple Silicon. The results are surprising: - K8V4 (8-bit keys, 4-bit values): 59% memory reduction with only 0.86% perplexity loss - K4V8 (4-bit keys, 8-bit values): 59% memory reduction but 6.06% perplexity loss - The configurations use the same number of bits, but K8V4 is 7× better for quality This means you can run LLMs with 2-3× longer context on the same Mac. Memory usage scales with sequence length, so savings compound as context grows. Implementation was straightforward: 1. Added --kvq-key and --kvq-val flags to llama.cpp 2. Applied existing quantization logic separately to K and V tensors 3. Validated with perplexity metric...

Thoughts on Thinking 43 by bradgessler | 15 comments on Hacker News.

The Scalar Select Anti-Pattern 5 by goranmoomin | 0 comments on Hacker News.

We Made CUDA Optimization Suck Less 7 by jaberjaber23 | 1 comments on Hacker News.

The Perverse Incentives of Vibe Coding 4 by laurex | 0 comments on Hacker News.

David Hilbert's radio address (2014) 10 by anigbrowl | 0 comments on Hacker News.

Dusk OS 79 by GTP | 37 comments on Hacker News.

FCC threatens EchoStar licenses for spectrum that SpaceX wants to use 22 by unsnap_biceps | 0 comments on Hacker News.

Build Your Own Siri. Locally. On-Device. No Cloud 5 by andreeamiclaus | 0 comments on Hacker News.

Toward a Sparse and Interpretable Audio Codec 8 by cochlear | 0 comments on Hacker News.

An online exhibition of pretty software bugs 14 by tobr | 0 comments on Hacker News.

Unique Games Conjecture 15 by surprisetalk | 1 comments on Hacker News.

Even Tesla's Insurance Arm Is Getting Wrecked 47 by ryan_j_naughton | 41 comments on Hacker News.

Odin, a Pragmatic C Alternative with a Go Flavour 17 by hmac1282 | 5 comments on Hacker News.

Inventing the Adventure Game (1984) 18 by CaesarA | 1 comments on Hacker News.

Show HN: Extension for full-text browser history search 18 by ApbNfMR | 8 comments on Hacker News. Hey. I’ve been working on Rearview, a browser extension that makes browsing history more useful. It’s available for Chrome, Edge, and Firefox. What it does: - Full-text Search: Search page content, not just URLs or titles. - More Filter: Organize history by date, time, or visits. - AI Assistant: Get insights from your history. (Optional, with your own api key.) Everything stays local in IndexedDB. Feedback & Feature request welcome. BTW, since Firefox does not provide a favicon URLs implementation ( https://ift.tt/RodF36B ),I am currently requesting icons in Firefox via https://icon.horse/ , the performance seems not ideal. Is there a more recommended solution?

Block Diffusion: Interpolating Autoregressive and Diffusion Language Models 6 by t55 | 0 comments on Hacker News.

Vagus Nerve Stimulation Erases PTSD 23 by rmason | 18 comments on Hacker News.

RSC for Astro Developers 3 by feross | 1 comments on Hacker News.

Will Supercapacitors Come to AI's Rescue? 24 by mfiguiere | 13 comments on Hacker News.

GenAI-Accelerated TLA+ Challenge 8 by lemmster | 1 comments on Hacker News.

Databricks in Talks to Acquire Startup Neon for About $1B 13 by ko_pivot | 7 comments on Hacker News.

Show HN: Real-time AI Voice Chat at ~500ms Latency 47 by koljab | 15 comments on Hacker News.

Show HN: Tkintergalactic - Declarative Tcl/Tk UI Library for Python 18 by leontrolski | 3 comments on Hacker News.

Reading Zanzibar 4 by surprisetalk | 0 comments on Hacker News.

Show HN: Free, 100% in browser PDF editor 52 by philjohnson | 9 comments on Hacker News. Add text, input boxes, pictures, signatures, delete pages, merge PDFs and password protect them. All happening in the browser, 100% free and no sign-up.

Elm Test Distributions 12 by todsacerdoti | 0 comments on Hacker News.

Hybrid AC/DC distribution system with a shared neutral (2020) 8 by 1970-01-01 | 0 comments on Hacker News.