pornel

5 ms·

1.
by pornel 6d ago
The big news here is they've replaced LLVM with MSVC's backend.
2.
by pornel 7d ago
Throwing it as-is into a simple bayesian filter works great too - nobody honest writes fun&zwsp;ding.
3.
by pornel 7d ago
Stellantis is still producing 400V EVs with below-average efficiency. Their cars display estimated range based on their marketing number rather than real battery condition and driving history, so the range drops by 1.6 miles for every mile
4.
by pornel 9d ago
> I fail to see how any sort of much lower level access to the machine would be portable across price ranges and microarchitecture generations That's the assumption that can be removed. GPUs don't have stable ISAs, and their as
5.
by pornel 9d ago
Modern CPUs hide so much logic that the assembly language is an abstraction itself. CPUs have many times more registers than their assembly language, execute instructions speculatively and out of order. Multiple layers of caches are synchro
6.
by pornel 12d ago
Apple would have probably lost more customers if it didn't also create a hard-to-leave lock-in ecosystem. Lots of people use old iPhones and don't care about some "up to" benchmark bumped every year, but are stuck with
7.
by pornel 14d ago
BTW, Rust's lifetime annotations for borrowed references are a mostly orthogonal feature. Liveness of objects for move/drop semantics is tracked differently, without any syntax and with implicit runtime drop flags where necessary.
8.
by pornel 14d ago
Not enough for coding. 48GB is minimum for a non-lobotomized coding model like qwen, and you'd likely want 64GB to have long context and not kernel panic when Chrome opens. You could run one of the smaller Gemma models to have a chatty
9.
by pornel 15d ago
There doesn't have to be any meaningful analogy here. Laws don't have to treat humans and machines equally. Laws are for people and allow things that are less than ideal, but necessary for people to function. Humans can't lea
10.
by pornel 15d ago
Unpleasant for interactive agentic work. Still useful to leave it to do some work in the background.
11.
by pornel 17d ago
The bug here is not related to virtualization, but a footgun as old as C stdlib: system() that doesn't take arguments separately, and instead relies on shell escaping by the application.
12.
by pornel 20d ago
Opus is good at deciding what to do beyond literal interpretation of the prompt (like checking assumptions before coding). Smaller models need to be told not only what to do, but how to behave. oh-my-pi's Advisor feature can fill a lot
13.
by pornel 20d ago
Because the other non-isolationist countries have the rest of the world to trade with. Their tariffs on USA don't make all of their imports more expensive, only redirect trade to other countries.
14.
by pornel 22d ago
You should be able to cancel it for free. In EU you also have 14 days to return products bought online, for any reason.
15.
by pornel 27d ago
You can take a working codebase, tell a shitty LLM to rewrite it badly, then swap it around to look like LLM-to-working code rewrite.
16.
by pornel 27d ago
You can't enforce those licenses against AI-generated parts, because enforcement relies on the recipient having no other way to avoid copyright infringement. However, there's no issue with including non-copyrightable code in oth
17.
by pornel 27d ago
Note that an incremental language improvement doesn't mean a small code change. For example, C++'s iterator with two related but technically independent pointers is a PITA for alias analysis, which is a problem not only for Circle
18.
by pornel 28d ago
1. In "verifiable" domains ML is not limited by training data any more. Models can help generate the training data and/or learn the objective through reinforcement learning. 2. Yes, because the most capable model is the only
19.
by pornel 28d ago
I expect this to be automated too, and in the end boil down to paying for more tokens to fix the program. Code too messy to be editable by an LLM is already too horrible for humans to touch. Fixing vibecoded software as a service will boil
20.
by pornel 28d ago
They use technical means to create lock-in that prevents me from NOT using these services. Apple actively prevents users from installing any other OS on an iPhone. Apple actively prevents uses from finding and installing apps through other
21.
by pornel 28d ago
> But the major players do seem to be happy to replace their C++ code with Rust. Because incremental improvements don't provide enough value. A stable C++ codebase is best left untouched. It's not worth to rewrite C++ into C+++
22.
by pornel 29d ago
> Aliexpress There are even cheaper ways of starting a fire.
23.
by pornel 1mo ago
Instead of prompting "Summarize: $SESSION" you submit "$SESSION Summarize what has been done so far". This way you have a prefix cache hit on the session and don't pay any more than for a regular prompt.
24.
by pornel 1mo ago
There is a solution that Mozilla uses in prod for legacy C codecs: https://rlbox.dev/ It's not a replacement for Fil-C's role as a precise ASAN/Valgrind, but it works great if you want to call a C library wit
25.
by pornel 1mo ago
I don't like that it throws away the whole KV cache when compacting. It costs a cache miss of the whole conversation length, and that's a waste of time and money. LLMs are perfectly capable of summarising the conversation without
26.
by pornel 1mo ago
Language models were supposed to be more flexible than Siri. Otherwise why bother with an LM? 14MB can fit a lot of regexes for fixed sentences.
27.
by pornel 1mo ago
Dictionary-based compression is based on prediction that recently seen words will be used again. That happens to be generally true for lots of datasets, including human languages (zipf distribution). JPEG's zig-zag is a primitive for q
28.
by pornel 1mo ago
Another example is encrypted data. Statistically, encrypted data is indistinguishable from random. Truly random data is impossible to compress losslessly. But if you had a predictor so smart that it could crack the encryption key, it could
29.
by pornel 1mo ago
You're just assuming a flawed implementation of a predictor and complain that it's flawed. That doesn't invalidate the general premise that predictors are compressors. Predictors don't have to give you some corpus-averag
30.
by pornel 1mo ago
Technical solutions may be seen as kindly as handcuffing yourself to something when told to leave. "I can't, I've made it impossible for myself to obey the law". They might not be able to break your private keys, but the

More ›