sylware

Yep, AI freedom of usage will only happen locally. Remotely, "That's all folks", The End.

Indeed, Linus T. is not superman, he cannot preserve linux of all the danger around.

doh!

(all that said, I don't know of anyone who would answer an unknown number call)

As far as I know, that was for x86(32bits) linux, that decades ago.

With those assembly source files (which do not abuse any pre-processor) and plain and simple C, I could build a modern x86_64 linux kernel with cproc/qbe (which gets 70% of gcc speed in my CPU intensive benchmarks... for a few % of gcc code and in plain and simple C, not brain damaged c++).

But I kind of don't mind since the future is assembly coding on non-IP-locked standard like RISC-V, and the main issue for that future is the abuse of pre-processors (ffmpeg was bitten by it) or code generators which would not be written in assembly themselves (or with a simple high level language with an assembly written interpreter, asmpython?).

Any success at transpiling it to C? Using the cfront transpiler improved with coding AI? :)

Yep, an ABI (function call convention) is computer language agnostic. It is a binary specification. And in real life, only a subset of it is actually used.

If they want to find something really obsolete, they better have a look at executable/dynamic lib file formats (PE+, ELF64). In other words, they better look at that first: I am using my own, which is beyond simple (a little RFC would suffice), no loader of any kind, basically userland syscalls. And I do embbed exes in an ELF64 capsule to run them transparently on linux systems (writting a internal linux exe loader would be copying ELF loading code while trashing 90% of its code). (hopefully in some not too far future, I'll try to build a mesa AMD vulkan driver for this very simple format and for that the main issue is.. c++ with its runtime, as always...).

Look at directx.

I skimmed the article, is this guy actually advocating for planned obsolescence or did I miss something??

"When other languages have compiler specific extensions beyond the spec, it is a failure in their design."

This one of the failures of Linus T. with the linux kernel: he was not able to keep the assembly source code with plain and simple C code you can compile with a small and alternative C compiler (same failure for the glibc devs I think).

I don't blame him, he is already keeping the linux ABI stable, and pulling that off is something.

Hasn't been already the case for a while now?

Oo

(and I am french)

What is generated is not machine code, but 'human' assemby to input into an assembler (was nasm). It is much much better.

Basically, if all that is really true, coding AIs may be our salvation from those abominations which are compilers.

Obsolete: it should be a binary specification with various implementations, even assembly.

spinof topic: android is still that java thingy? Or is it finally a set of modular binary specifications which are computer language agnostic?

Spam relies on trust the user gives to the domain name in the email address of the from header, you dumb dumb.

There are doing exactly that... slowly and surely... pure evil.

Namely AMD would have to publish the hardware documentation (for instance command packets) of anything specific to inference (unless, this is just their GPU doing only compute).

Your rumors have one less 0 than my rumors. :)

I have no clue anyway.

How???

Allright! Well, the rumors I got are around 100T for those frontier models: reading stuff here and there on internet, on discussion forums with guys pretending running AI models, etc.

This is so hard to sort the true from the false nowadays.

If LLM becomes that good at coding, I'll have to run an open weight one locally.

My question is that wrong?

I would prefer direct hardware kernel interface.

Like linux DMABUFs with userland hardware command ring buffers (I guess this hardware ring buffer instance would be specific to a VMID and a PASID).

Rumors say chatgpt/claude/gemini/etc are in the 100s of teras. True?

Is their hardware programming interface reasonable for implementing inference of frontier models: no quantization, several tera params?

BTW, how many many params open weight frontier models have? A few teras, 100s of teras?

Regulation on small tech interop is exactly for guys like you.

Open source is not enough: _LEAN_ open source is a strong requirement today... and that includes syntax complexity of any used computer languages.

The scammers are now doing open source vendor/dev lock-in via massive size and complexity... while there are small and stable in time alternatives able to do a good enough job. Those guys are not all malicious, many are just brain washed and some suffer from a mental disorder: accute Rube Goldberg Machine syndrome.

Spam does not work with email addresses using IP literals.

twitch.tv has new HLS streamers which break ffmpeg HLS.

Use lynx, links or netsurf. Or vibe code your noscript/basic HTML web engine?

I think DMARC is missing email address with IPv[46] literals support. As being self-hosted, without paying the DNS mob, I am still blocked to send email to gmail.com because such email addresses do throw out of whack gogol code.

Email addresses with IPv[46] literals are intrinsincly stronger than SPF. If in the envelope or any of the 'from' headers (if my memory does not fail me, there are few more headers to scan), the IPv[46] literal does not match the actual and real IP of the SMTP server, the email is dropped, not even going into any spam folder.

Conspiracy mode: they know and are careful not to support that, in order to create a walled garden of internet messaging for them and their friends.

Weird.

I download RSS from gogol: youtube channel information. Why, because the real evil of gogol and its friends: the whatwg cartel. Their web engines are technical abominations they pushed down the throat of the whole planet, because they actually control all (including the SDK then the computer language). How? They break on purpose interop with small tech web engines, aka noscript/basic HTML.

What's really annoying are major authorities unable to regulate that, namely force interop with small tech (network protocols/file formats)... like they know but are careful not to do anything with a significant impact. Well, In my EU country, this is decided at the top level: prime minister and president and that's it: this is a sorta proof that matter is actually really important. Everything started to go wrong from my EU country gov of 2015/2016 since they 'forgot'(ooopsie!) small tech web engines in the 'master document'.

There is still hope, once small tech is added again to this 'master document' (web engines, currated subset of network protocols and file formats), interop with small tech web engines (most are EU) will have to be added, in 3 years.