gapry @_gapry

Parsers don’t have to be complicated bkaradzic.github.io/posts/scanner/

For my first post, I’m sharing a letter @NVIDIA signed on why open models matter. AI will transform every industry, power every company, and be built by every country. Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty. The world needs both frontier closed models and frontier open models. images.nvidia.com/pdf/Open-Weigh…

keyboard_arrow_left Previous keyboard_arrow_right Next

[ #compiler ] I wrote a study notes blog post on Day 11 of Matt Godbolt's Advent of Compiler Optimisations 2025 series: "Counting Bits". Read more here: gapry.github.io/2026/05/31/Adv…

GPT-5.5-medium has lower end-to-end latency, uses less tokens and is overall smarter and cheaper than Gemini 3.5 Flash it might genuinely be over for anyone not named OpenAI or Anthropic

Lisan al Gaib @scaling01

2 months ago

some more Gemini 3.5 Flash benchmarks by Artificial Analysis AI: - the APEX-Agents-AA score is excellent - expected higher on CritPt - reasoning efficiency could also be better, but it kind of depends on what setting they used. if it's the max then it's very good - Price/Perf

keyboard_arrow_left Previous keyboard_arrow_right Next

6 15 204 186K 35

2026 is 32% complete.

[ #compiler ] I wrote a study notes blog post on Day 10 of Matt Godbolt's Advent of Compiler Optimisations 2025 series: "Unrolling loops". Read more here: gapry.github.io/2026/04/30/Adv…

TIL the demoscene (which heavily used hand written assembly, long before FFmpeg did) is a form of UNESCO intangible cultural heritage demoscene-the-art-of-coding.net

Robert Clausecker @FUZxxl

4 months ago

@thr0wsException @FFmpeg It's called the demoscene and it's recognised by UNESCO as an intangible cultural heritage in Germany and Finland.

1 6 112 45K 15

Introducing MiniDXNN: a new MLP library built for #DirectX12 HLSL, enabling neural rendering without compute API interop: 🧠 Optimized MLP inference kernels. ⚡ Acceleration via cooperative vector APIs. 🎮 Sample apps & full source code. Get started ⬇️ gpuopen.com/learn/minidxnn…

Merkle trees are everywhere: - ZFS uses them to detect data corruption - Git uses them to verify repo integrity - Cursor uses them for codebase sync - Bitcoin uses them for transaction verification Talked through how they work on the latest database stream.

MAD Bugs: Claude Wrote a Full FreeBSD Remote Kernel RCE with Root Shell (CVE-2026-4747) To our knowledge, this is the first remote kernel exploit both discovered and exploited by an AI. blog.calif.io/p/mad-bugs-cla…

Finally, with Clang 22's release, here's v1 of Linux kernel slab cache partitioning with help of Clang's AllocToken feature: lore.kernel.org/all/2026033111…

An automated N-day research pipeline at PwnFuzz. Ghidra + Ollama + n8n →Diffs Patch Tuesday binaries → LLM analyzes the output → Structured vuln reports, monthly AI-generated reports gets you oriented fast! Blog: ghostbyt3.github.io/blog/nday-rese… Repo: github.com/ghostbyt3/nday…

The FLARE team now freely distributes its quality reverse engineering and malware analysis educational content at github.com/mandiant/flare…. Launched with: - Malware Analysis Crash Course - Go Reversing Reference - Intro to TTD

Samsung used RISC-V in one of their consumer commercial products for the first time. For the BM9K1, a PCIe 5.0 QLC SSD with 11.4GB/s speeds coming in 2027, it used RISC-V cores instead of the usual ARM Cortex-R in the controller.

2026 is 26% complete.

LLM Knowledge Bases Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So: Data ingest: I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them. IDE: I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides). Q&A: Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale. Output: Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base. Linting: I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into. Extra tools: I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries. Further explorations: As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows. TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

[ #compiler ] I wrote a study notes blog post on Day 9 of Matt Godbolt's Advent of Compiler Optimisations 2025 series: "Induction variables and loops". Read more here: gapry.github.io/2026/03/31/Adv…

[ #compiler ] I wrote a study notes blog post on Day 8 of Matt Godbolt's Advent of Compiler Optimisations 2025 series: "Going loopy". Read more here: gapry.github.io/2026/03/31/Adv…

[ #compiler ] I wrote a study notes blog post on Day 7 of Matt Godbolt's Advent of Compiler Optimisations 2025 series: "Division by a Constant". Read more here: gapry.github.io/2026/03/31/Adv…

@soyjuanarbol @golang Why not both?