Hacker News Reader

codexonon Meta’s renewed commitment to jemalloc

Agreed mostly. Going from standard library to something like jemalloc or tcmalloc will give you around 5-10% wins which can be significant, but the difference between those generic allocators seem small. I just made a slab allocator recently for a custom data type and got speedups of 100% over malloc.

codexonon Meta’s renewed commitment to jemalloc

> It was mainly pushed as easy to adopt, easy to use, easy to statically link, etc.

That is true of basically every single malloc replacement out there, that is not a uniquely defining feature.

codexonon Meta’s renewed commitment to jemalloc

I've benchmarked them every few years, they never seem to differ by more than a few percent, and jemalloc seems to fragment and leak the least for processes running for months.

Mimalloc made the claim that they were the fastest/best when they released and that didn't hold up to real world testing, so I am not inclined to trust it now.

codexonon Meta’s renewed commitment to jemalloc

It still beat mimalloc when I checked 4-5 years ago.

codexonon Meta’s renewed commitment to jemalloc

I've been using jemalloc for over 10 years and don't really see a need for it to be updated. It always holds up in benchmarks against any new flavor of the month malloc that comes out.

Last time I checked mimalloc which was admittedly a while ago, probably 5 years, it was noticebly worse and I saw a lot of people on their github issues agreeing with me so I just never looked at it again.

codexonon AI Added 'Basically Zero' to US Economic Growth Last Year, Goldman Sachs Says

I'm not arguing about growth. I was addressing this statement which seems to presume that AI has no effect if the job can't be removed.

> If AI can't do 100% of a job then you can't remove the job.

codexonon AI Added 'Basically Zero' to US Economic Growth Last Year, Goldman Sachs Says

Claude code will prompt you and explain to you what practice fits a situation. It might not do it perfectly, but the foundations are there.

codexonon AI Added 'Basically Zero' to US Economic Growth Last Year, Goldman Sachs Says

AI is already aware of the best practices. It does not just blindly do what you ask of it in the simplest way.

codexonon AI Added 'Basically Zero' to US Economic Growth Last Year, Goldman Sachs Says

I'm not sure that's entirely true. For most things, checking if a solution is correct is much easier than implementing it (page looks wrong, can't login etc...)

codexonon AI Added 'Basically Zero' to US Economic Growth Last Year, Goldman Sachs Says

You can replace it with a much lower paid employee though.

codexonon OpenClaw is changing my life

I think some of it might be genuine. For people that don't code (like management), going from 0 to being able to create a landing page that looks like it came from a big corporation is a miracle.

They are not able to comprehend that for anything more complicated than that, the code might compile, but the logical errors and failure to implement the specs start piling up.

codexonon Top AI models fail at >96% of tasks

Check the link to the study. It has been updated for Opus 4.5.

codexonon Top AI models fail at >96% of tasks

This paper creates a new benchmark comprised of real remote work tasks sourced from the remote working website Upwork. The best commercial LLMs like Opus, GPT, Gemini, and Grok were tested.

Models released a few days ago, Opus 4.6 and GPT 5.3, haven't been tested yet, but given the performance on other micro-benchmarks, they will probably not be much different on this benchmark.

codexonon Claude Opus 4.6

https://arxiv.org/abs/2510.26787

Testing the top llms on wework, the highest performing one only succeeded with a rate of 2.5%

Can you imagine not being fired when you can only do 2.5% of all tasks?

This study is dated October 30th, very recent.

codexonon Claude Opus 4.6

You don't need to be a genius or rocket scientist to write code, but llm don't even reach the bar for anything but the most simple things. Take a look at the video I posted earlier for an example.

And specialised models for programming HAVE plateaued.

https://livebench.ai/#/?sort=Agentic+Coding+Average

From Claude 4.1 to 4.5 was only an 18% gain, and from 4.5 to 4.6 it even DECLINED. Codex 5.1 to 5.2 also shows a decline.

codexonon Claude Opus 4.6

Top AI researchers like Yann LeCunn have said that LLMs are a dead end.

It seems to me that LLM performance is plateuing and not improving exponentially anymore. This recent hubbub about rewriting a worse GCC for $20,000 is another example of overhype and regurgitating training data.

You don't know for sure if it is going to "snow" (AI reaches general intelligence) Snow happens frequently, AI reaching general intelligence has never happened. If it ever happens, 99% of jobs are gone and there is really nothing you can do to prepare for this other than maybe buy guns and ammo, and even that might not do anything to robotic soldiers.

People were worried about AI taking their jobs 60 years ago when perceptrons came out, and anyone who avoided a tech career because of that back then would have lost out majorly.

codexonon Claude Opus 4.6

Even for coding, it seems to still make A LOT of mistakes.

https://youtu.be/8brENzmq1pE?t=1544

I feel like everyone is counting chickens before they hatch here with all the doomsday predictions and extrapolating LLM capability into infinity.

People that seem to overhype this seem to either be non-technical or are just making landing pages.

codexonon Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model

Didn't this paper demonstrate that you only need 1.58 bits to be equivalent to 16 bits in performance?

https://arxiv.org/abs/2402.17764

codexonon The microstructure of wealth transfer in prediction markets

Has anyone noticed a lot of polymarket posts on their X (formerly known as twitter) feed claiming to be making a fortune? It makes me feel like its some kind of coordinated guerilla marketing scheme.

codexonon AI generated music barred from Bandcamp

Recent update proving my point

https://www.notebookcheck.net/Steam-updates-AI-disclosure-fo...

codexonon AI generated music barred from Bandcamp

That's not really the same as stores outright banning AI code.

An apt analogy would be like a shared drawing taking merge requests and having to spend 30 minutes looking at every single merge request zoomed in to see if there was a microscopic phallus embedded somewhere.

It is completely fair for an open source project to have their own standards, and you are also free to fork it so you can accept as many AI PRs as you want.

None of these options are available for someone that wants to sell AI generated music. There are really only 2 marketplaces to sell your own music and if both of them banned AI, then you are effectively locked out of the entire market.

codexonon AI generated music barred from Bandcamp

They are obviously different things, but aren't the people who spent thousands of hours honing their coding and releasing their code spending just as much time and effort if not more than the people who made non-ai images and music?

codexonon AI generated music barred from Bandcamp

Yes and it was probably only done because of people complaining about AI art, not AI code.

codexonon AI generated music barred from Bandcamp

AI music from suno sounds indistinguishable to non-ai generated music to me.

In terms of how well it works, the quality of AI music is far better than art or code. In art there are noticeble glitches like multiple fingers. For code, it can call non existent functions, not do what it is supposed to do, or have security issues or memory leaks. From what I can tell, there is no such deal breaker for AI music.

codexonon AI generated music barred from Bandcamp

Sure, I am one of the people who will say that. But where are the people calling for it to be banned? Where are the stores and websites that are banning AI generated software?

codexonon AI generated music barred from Bandcamp

I find it interesting that there's so much pushback against ai generated art and music while there seems to be very little for ai generated code.

codexonon A closer look at a BGP anomaly in Venezuela

Lower cost usually means lower quality and is an example of how a long path being leaked can result in traffic flowing away from high quality path to the leaked path.

Not saying that this is the case with Venezuela, just explaining the reality of BGP where path prepends are often ignored.

codexonon A closer look at a BGP anomaly in Venezuela

Actually many networks will prefer routing over a cheap AS path no matter how long it is.

codexonon Show HN: I made a down detector for down detector

I've actually tried hetzner on and off with 1 server for the past 2 years and keep running into downtime every few months.

First I used an ex101 with an i9-13900. Within a week it just froze. It could not be reset remotely. Nothing in kern.log. Support offered no solution but a hard reboot. No mention of what might be wrong other than user error.

A few months later, one of the drives just disconnects from raid by itself. It took support 1 hour to respond and they said they found no issue so it must be my fault.

Then I changed to a ryzen based server and it also mysteriously had problems like this. Again the support blamed the user.

It was only after I cancelled the server and several months later that I see this so I know it isn't just me.

https://docs.hetzner.com/robot/dedicated-server/general-info...

codexonon Azure hit by 15 Tbps DDoS attack using 500k IP addresses

Why wouldn't microsoft advertise this though? If they had the ability to take the attack and others might not, then it'll result in more customers for them.