You are presuming an existing tile IP - if you're already in possession of such an IP then the place and route is already coarse grained. There are lots of papers on this.
> (like the GPU does)
What exactly does the GPU do? Yes there are tiles but it's up to you to now tile your workload. You understand this is the exact same problem you're bemoaning re place and route - you need to figure out how to shuffle individual bits efficiently through an existing fabric (roughly it's the same thing as routing 32 wires at a time).
> we need for this to happen is for the companies making the FPGAs to open up the board layout file spec
What exactly is this going to do for you if you're placing tiles? Also you can already recover this by exhaustively enumerating all a->b paths (yes people really do this).
To anyone else that thinks they just absolutely are certain of the silver bullet for digital design: download a copy of Vivado and report back what you discover!
> The time it needs to run is irrelevant for its correctness. And so they can stack and stack and stack
This is a very naive take - the very direct translation of what you're saying doesn't happen does in happen in analysis all the time: there are many inequalities which can be "stacked" to prove a bound on something but their factors are too large so you cannot just stack them if you need a fixed bound for your proof to go through. Unsurprisingly this is exactly how actual runtime analysis also works (it's unsurprising because they're both literally math).
I think you are taking it a bit out of context here. Obviously, I assumed in that phrase, that mathematicians are stacking suitable methods and proofs. If some factors are too large to not fit in some bound, then obviously that's not something you would stack. But once you have suitable proofs and proven correct and suitable methods, you can stack, and correctness does not go out of the window. Correctness remaining, you will be able to get to a proven correct result.
Of course proving things in mathematics is also a lot harder, usually, than computer programming, and it is probably still easy to make mistakes.
Well after a certain point people have to smell the roses, so to speak. You don't get to control your business 100%, the market tells you what to do a lot of the time.
I think, the reality is, as models become more competitive they are becoming commodities. There's really no reason an app has to be built on GPT, or Gemini. It makes much more sense for apps to be "model agnostic" and let their customers choose which models to use.
I think, if OpenAI sticks to just trying to make their own apps for everything, they will be outrun. People will make apps outside of their ecosystem and will just use them as an API dumb pipe, regardless of if OpenAI wants that. And if they don't want that and restrict it, then their models will fall to the wayside as more competitive models which DO allow that take their place.
They're in a bind here, which is probably why we are seeing this announcement. OpenAI can see the writing on the wall for them.
No it's not - the compiler for MSL is of course C++ because it's LLVM but the runtime is absolutely written in objc (there weren't even C++ bindings until recently).
No, I mean what is inside the Objective-C objects. Essentially everything on macOS has an Objective-C API but is implemented using C++. Have you ever noticed the ".cxx_destruct" method on like all objects?
What you are talking about are C++ wrappers around Metal Objective-C API. Yes, it is weird as they are going C++ -> Objective-C -> C++. Why not go directly? Because Apple does not ship C++ systems frameworks.
The commenter you’re replying to, in their heart of hearts, truly believes in 5 years that an LLM will be writing the majority of the code for a project like say Postgres or Linux.
Worth bearing in mind the boosters said this 5 years ago, and will say this in 5 years time.
Everyone working in programming is writing code for a project more like Postgres or Linux than they are a project like making a wood cabinet or a life drawing.
> Perfectly good excuse to make society worse for people
What an incredibly silly accusation to make of a company/service that streams movies and television. Like you understand it is possible to dilute the concept of civic responsibility right?
Not really... A true concurrent high-throughput low-latency garbage collector is the stuff of nightmares. Just ask Azul Systems that sells real-time JDK for the high-performance trading applications.
IREE hasn't been at G for >2 years.