Hacker News Reader

yukIttEfton Microsoft patches Windows and Excel – breaks audio, remote access, and paste

There is not a single component in Windows that is not riddled with bugs, its infuriating.

I wonder what exactly the "quality czar" is doing, maybe he needs a good PIP?

https://www.theregister.com/software/2026/02/05/satya-nadell...

What are your purposes?

I'm the same. What helps me is Blender MCP. Then you can ask it how to do things and it will either do it or explain how to do it.

https://www.blender.org/lab/mcp-server/

yukIttEfton Unauthenticated RCE in Motorola's MR2600 Router

In Germany we have "Störerhaftung" where routerowners are responsible for everything that happens through their router.

I wonder how this would hold up in court, couldn't you argue that routers are generally buggy, how can they force any responsibility if they can easily be hacked?

yukIttEfton Rewriting Bun in Rust

Is it though.

There are so many situations where something is guaranteed to be safe but there is no way to express that in the Rust typesystem, so the only thing you can do is to wrap everything in Arcs and Mutexes, which introduces allocations, pointerchasing and locks

yukIttEfton Grok 4.5, based on our 1.5T V9 foundation model, with Cursor data added in su

Can the gas turbines be stopped then?

https://www.wsj.com/pro/sustainable-business/doj-defends-mus...

yukIttEfton How LLMs work

> so the model figures out during training what each token should look for and what it should offer

But how does it learn this token-relationship?

All it has is many text samples, but still, nowhere it says how the tokens relate to each other, so where does this information come from?

yukIttEfton 80386 microcode disassembled

If you put this into an emulator, would it boot linux?

yukIttEfton Anthropic downgraded cache TTL on March 6th

I'm currently "working" on a toy 3d Vulkan Physx thingy. It has a simple raycast vehicle and I'm trying to replace it with the PhysX5 built in one (https://nvidia-omniverse.github.io/PhysX/physx/5.6.1/docs/Ve...)

I point it to example snippets and webdocumentation but the code it gens won't work at all, not even close

Opus4.6 is a tiny bit less wrong than Codex 5.4 xhigh, but still pretty useless.

So, after reading all the success stories here and everywhere, I'm wondering if I'm holding it wrong or if it just can't solve everything yet.

yukIttEfton Tinybox – A powerful computer for deep learning

He had a video on Youtube where he proudly gloated about how he voted for Trump in not one but two elections, how happy he is that he can now openly talk about it, how its a fresh start for US, how catastrophic Harris would have been.

Did he take down the video because of embarrassment or did he fear negative impact on his sales?

yukIttEfton Microsoft account bugs locked me out of Notepad – Are thin clients ruining PCs?

Tried switching to Linux years ago, what brought me be back were issues with Suspend-to-disk, CPU Core parking, GPU powerefficient idle mode

Since my PC is idle most of the time, I care about idle-powerefficiency a lot.

Does stuff like this these days just work? (AMD CPU + Nvidia GPU)

yukIttEfton Language models pack billions of concepts into 12k dimensions

newbie question: when training networks, what mechanism makes the language's concepts be (almost)orthogonal to each other?

yukIttEfton The Effect of Noise on Sleep

I have a street in front of my bedroom and I'm wondering how well ANC earbuds work in that case. Will it be completely silent or will there still be some kind of whoosh-sound?

yukIttEfton Denuvo Analysis

Wasn't aware of Alive2. Thx, I'll have look.

I had some success with https://github.com/mrphrazer/msynth But its hard to glue this to LLVM.

yukIttEfton Denuvo Analysis

Had also a look at Denuvo a while ago. Used LLVM to remove the x86 obfuscation and broke it down to VM-Opcodes. Atleast back then, Denuvo seemd to translate gamecode into a stackmachine.

This is how a VM push looks like:

      temp[0]=add(mem[e268], fffffffffffffff8)
      mem[temp[0]]]=mem[e560]
      mem[e268]=temp[0]

(vmreg_e268 is stackpointer, its decremented and stored in tempreg, then the value of vmreg_e560 is copied to stackpointeraddr, then new stackpointervalue is written back)

But i quickly lost interest when it became MBA galore:

      temp[7]=sub(add(add(and(mem[ebe8], b2f7), 3fd8), xor(lshr(mem[ebe8], 1), 2684)), lshr(add(mem[ebe8], b2f8), 1))
      temp[d]=or(sub(sub(4ad, temp[7]), xor(and(shl(temp[7], 1), 95c), 95c)), 8000)
      temp[e]=lshr(temp[d], 1)
      temp[11]=lshr(add(temp[d], 8001), 1)
      mem[ebe8]=sub(xor(xor(temp[e], 3fff), temp[11]), shl(and(and(temp[e], 3fff), temp[11]), 1))

(looks like its doing some operation with a constant to vmreg_ebe8, but obfuscated by MBAs and most likely the result won't ever being used, so its just noise to drown out the real operations)

BTW: anyone aware of LLVM optimizer passimplementations that can deal with MBAs ?

yukIttEfton Reverse engineering the 386 processor's prefetch queue circuitry

When are you going to implement the first electron-level 386 emulator?

yukIttEfton Material Theme has been pulled from VS Code's marketplace

> We do not plan to add a permission model in the next 6 months.

I guess Copilot functionality trumps "Security above all else" now.

https://blogs.microsoft.com/blog/2024/05/03/prioritizing-sec...

yukIttEfton Notes on OpenAI's new o1 chain-of-thought models

Makes me wonder if "I don't know" could be added to LLM: whenever an activation has no clear winner value (layman here), couldn't this indicate low response quality?

yukIttEfton Intel SGX Fuse Key0, a.k.a. Root Provisioning Key Was Extracted by Researchers

I thought microcode was encrypted, how do they have its disasm? https://xcancel.com/pic/orig/media%2FGV7BLfeXsAEJ6HC.png

yukIttEfton AMD's "Peano" – An LLVM Compiler for Ryzen AI NPUs

Generally speaking, AI Engine processors are in-order, exposed-pipeline VLIW processors. Each VLIW instruction bundle specifies the behavior of one or more functional units, which begin executing a new instruction at the same time. The processor pipeline does not include stall logic to enforce data dependencies, and instructions will continue executing in order regardless of other instructions in the pipeline. As a result, the compiler is able to schedule machine instructions which access the same register in ways that potentially overlap

So a software scheduler? I had no idea it works like this. Very interesting.

yukIttEfton C++ Insights – See your source code with the eyes of a compiler

This is very helpful, thx.

yukIttEfton C++ Insights – See your source code with the eyes of a compiler

I wish there was a way to show the code that is generated when using co_return or co_await

e.g. https://cppinsights.io/lnk?code=Ly8gaHR0cHM6Ly93d3cuc2NzLnN0...

yukIttEfton How does Single Sign-On (SSO) work?

newbie question: in https://substackcdn.com/image/fetch/f_auto,q_auto:good,fl_pr... What is the point of step 8/9? Couldn't a wrong password already be rejected in step 5/6?

yukIttEfton Intel Gaudi2 chips outperform Nvidia H100 on diffusion transformers

I'm wondering how AI scientists work these days. Do they really hack Cudakernels or do they plug models together with highlevel toolkits like pytorch?

Considering its the latter, considering pytorch takes care of providing optimized backends for various hardwares, how big of a moat is Cuda then really?

yukIttEfton Comments Are Code (2018)

I like to think that asserts are a form of active documentation

yukIttEfton Microsoft actions following attack by nation state actor Midnight Blizzard

From the same company that charges you to access your logfiles.

https://www.theregister.com/2023/07/20/under_cisa_spressures...

yukIttEfton Samsung extends Android and security updates to 7 years

I mean, do they even have to do more than run "make image" after rebasing to a new (minor) kernel update ?

yukIttEfton Apple built iCloud to store billions of databases

Would you mind sharing what made you quit Snowflake? (I'm considering applying there)

yukIttEfton AMD MI300 performance – Faster than H100, but how much?

H100 GPU is a sibling of NVidia's customer RTX 40-series: https://en.wikipedia.org/wiki/Hopper_(microarchitecture)

Is MI300 also a variant of their customer GPUs?