I wasted a bunch of time trying to figure out why xterm starts so slowly on Windows... and solved it with a dumb LD_PRELOAD hack 😆 lock.cmpxchg8b.com/slowterm.html

CVE-2024-37085 #VMWare

Alan Gray (NVIDIA) writes about implementing CUDA graphs in llama.cpp: developer.nvidia.com/blog/optimizin…

Today I joined a very exclusive club. The club of people who have been dragged off stage at DEF CON from the talk there were meant to participate in. Thanks to all who then came out to my outside talk. Glad I was able to answer all question. It was fun.

Let's bring llama.cpp to the clouds! You can now run llama.cpp-powered inference endpoints through Hugging Face with just a few clicks. Simply select a GGUF model, pick your cloud provider (AWS, Azure, GCP), a suitable node GPU/CPU and you are good to go. For more info, check

Уърдъл 🇧🇬 1000 5/6* ⬜⬜🟨⬜⬜ ⬜🟩⬜🟩⬜ ⬜🟩⬜🟩⬜ ⬜🟩⬜🟩🟨 🟩🟩🟩🟩🟩

Getting made fun of because you cover laptop webcam with a sticker? 😭 Here are materials from my talk about controlling ThinkPad X230 webcam LED over USB presented at POC by POC_Crew 👨‍👩‍👦‍👦 😎 Use these as a comeback 😁 Slides: docs.google.com/presentation/d… Code: github.com/xairy/lights-o…

Reversing MAGic Remote: xakcop.com/post/magic-rem…

Never use default Wi-Fi passwords: xakcop.com/post/smartcom-…

We also used to make phone cards with PIC16F84 until they adopted challenge-response authentication. I miss the nineties 🥲

[#Zer0Con2025] - SPEAKER 9️⃣ 💁‍♂️Ruikai - Hardcore Inference Attack: Unraveling Llama.cpp’s RPC Heap Puzzle

14 years later

Retweet if you want llama.cpp added here 👇

this is so insane. kCTF has a first-come-first-serve policy when it comes to 0day bounties when an instance releases. this team hand crafted a proof of work solver with avx-512 instructions to beat everyone else with an 0day to the flag: anemato.de/blog/kctf-vdf

github.com/ggml-org/llama…

🚀🚀🚀

A few highlights: 1.18 Tflops cluster FP64 on CPU (308 Gflops single node) DeepSeek R1 671B Q2_K - 8.25 t/s on cluster gpt-oss 20b - 45 t/s on single node gpt-oss 120b - 33 t/s on single node gpt-oss 120b - 24 t/s on cluster (llama.cpp RPC)

> copy other people's homework > claim that they made it themself classic move, ollama

whisper.cpp is coming to ffmpeg github.com/FFmpeg/FFmpeg/…