23,650 karma · joined October 15, 2013
Forth might work better than C, but tbh, assembly code is more straightforward and easier to read than Forth ;)
Also, looking at the code, the whole project looks vibe coded, no human would write such comments. Everything in a single git commit also looks "suspicious". It's cool that LLMs can write Z80 asm now, but yeah "labour of love"... tsk tsk tsk...
LLMs are an additional way to generate code, but it's not like code generation or "low-code" tools are a new thing that made compilers and programming languages obsolete.
LLMs "just" allow to provide the code generation input directly in human language. But whether providing such a specification in "sloppy" human language instead of a precise specification language is actually a good thing for long term projects is still arguable (as convenient as it is).
Mathematicians and engineers invented their own precise "DSLs" (e.g. math notation and blueprints) for exactly this reason, and high-level programming languages are nothing else than "specification languages" that don't leave room for the ambiguities of "natural languages", and it's still not clear to me how having both a natural language specification and a precise specification in form of source code side by side can be considered a good thing. It's easy to say of course that the "natural language specification" is the single source of truth and the source code is a throwaway output artifact, but then you're back at having a specification written in a language that's useless for expressing precise intent.
(and this is also the fundamental difference between compilers and LLMs and why I believe that comparing LLMs to compilers is wrong: determinism. The output artifacts of a compiler are literally throw away. Nobody in their right mind would commit object files into version control, while nobody in their right mind would not commit the source files generated by LLMs into version control, because creating the source files from scratch would yield different results on each run).
Releasing the harness as part of LLM-generated projects still makes sense of course :)
E.g. floating point behaviour is only the tip of the iceberg. Every GPU vendor also does mipmapping differently (to balance image quality versus required memory bandwidth).
Everybody else can write such tools now by themselves in at most an hour, and they get a tool that's more personalized to their tastes. And why have a readme at all when it's just LLM mumbo-jumbo - that sort of mumbo-jumbo is written for LLMs to consume, not humans.
On the contrary, WebGL2 is probably the only 3D API that comes even close to the idea of "write once, run everywhere".
TL;DR: don't expect that floating point operations on GPUs are strictly IEEE-754 compatible
For starters, neither Israel nor Azerbaijan threaten Europe with nuklear annihilation on a nearly weekly basis.
Lol, ok wacko... whatever you say.
Otherwise macOS 27 is mostly okay-ish though, especially compared to 26.
Remember: it was Russia which tried to blackmail Germany by shutting down the pipelines, and that was before the Ukrainians settled the matter once and for all by blowing up Nordstream (Russia had been shutting down the gas flow indefinitely by end of August 2022 after three months of back-and-forth, and the dry pipeline was finally blown up end of September 2022).
...of course you're likely just trolling, hard to say with the drivel coming out of Silicon Valley the last few years. In that case: my apologies and keep up the good work :D
In a way it's a return to waterfall, just with faster implementation phases.
Also related fun fact: the GDR started to refurbish and use old steam locomotives again as a result of the oil crisis. I remembered seeing those regularly as a child, and then later wondered what caused this anachronism.
It all depends of course where the code sits between "low level mission critical code" and "high level throw-away code". every non-trivial project has code across that entire range.
E.g. the lower level the code, the more I want to get directly involved, and that may mean typing each line manually to get that code directly into my head (IME for that purpose, writing is better than reading - have that manually written code reviewed by LLM instead).
Of course that only works if you don't have a manager breathing down your neck who insists on prioritising implementation speed over quality ;)
Also you should try assembly coding with modern (non-LLM) tooling, it feels incredibly more productive than with the primitive tools on slow machines of the past.
Also:
> software engineers can focus on data structures, software architecture and algorithms.
Come on, this is was already "normal" decades before LLMs, that's why an entire zoo of high level languages was created which allow to describe a solution in a highly abstract (yet still precise) way. Replacing this high level problem specification in a programming language with a "sloppy" human language specification is arguably even a step back.
(also tbf, I'm writing this comment while sitting in a cafe while the LLM at home is busy coding a GL mocking library ;) (which is exactly the boring type of throwaway code which should be automated)
...OTH if Britain wouldn't have been a globe-spanning empire, they wouldn't have needed a more precise way to measure longitude, and probably wouldn't have invented it ;)
E.g. the Brits had built a whole government body around the task, which then resulted in a competition, with Harrison's chronometer coming out as winner: