There is another big change in gpt-4o-2024-08-06: It supports 16k output tokens compared to 4k before. I think it was only available in beta before. So gpt-4o-2024-08-06 actually brings three changes. Pretty significant for API users
1. Reliable structured outputs
2. Reduced costs by 50% for input, 33% for output
3. Up to 16k output tokens compared to 4k
I often make fun of McKinsey- style four quadrants when overused, but they really boil down to something that makes a lot of sense in communicating a problem space:
a) carefully choose the two most important dimensions of concern (as Alan Kay said: the correct point of view is worth 80 Iq points)
b) make them binary: are we happy here or do we need to change?
In a way similar to the pareto ratio, you keep a surprising amount of value in something “so simple it cant be possibly so useful”.
Luckily there are faster and smaller alternatives in Rust for the ElasticSearch - Toshi[1], Meili[2] and Sonic[3]. In the age of Rust there is no need to use JVMs overhead.
Reframe it in your mind. "Getting into FPGAs" needs to be broken down. There are so many subsets of skills within the field that you need to level expectations. No one expects a software engineer to jump into things by building a full computer from first principles, writing an instruction set architecture, understanding machine code, converting that to assembly, and then developing a programming language so that they can write a bit of Python code to build an application. You start from the top and work your way down the stack.
If you abstract away the complexities and focus on building a system using some pre-built IP, FPGA design is pretty easy. I always point people to something like MATLAB, so they can create some initial applications using HDL Coder on a DevKit with a Reference design. Otherwise, there's the massive overhead of learning digital computing architecture, Verilog, timing, transceivers/IO, pin planning, Quartus/Vivado, simulation/verification, embedded systems, etc.
In short, start with some system-level design. Take some plug-and-play IP, learn how to hook together at the top level, and insert that module into a prebuilt reference design. Eventually, peel back the layers to reveal the complexity underneath.
I'm a big fan of this genre of game, I have a small but growing steam collection of different ones. ( And some non-steam such as nandgame ).
Turing complete is one of the better ones, the simulation seems reliable and the missions, after a fair bit of early access re-jigging are reasonably well structured.
However, the UI isn't always super smooth, and the lack of hand-holding means it's better for someone like me who's gone through a similar journey before in other simulators.
If you're looking for something similar that does a better job at explaining the concepts then I'd recommend "Silicon Zeroes". The game as a whole feels even slightly less polished, but has a cute storyline and in my opinion does a better job at explaining each level and what it's trying to introduce, and does a much better job with problem solving issues related to "The Clock".
If you don't actually enjoy the the electronics part so much and just enjoy the psuedo-assembly aspect, then I'd recommend TIS100 or Shenzhen I/O.
If you prefer well-crafted puzzles then something like "Human Resource Machine" may be more your style.
Fyi, if you are ever looking for a fun project you might be able to implement this. The vscode editor source is available as a library https://microsoft.github.io/monaco-editor/
I think, because the reason is what we want from our tools is different. I already use a much sophisticated tool for my tasks, and project planning (Pagico), and what I want is a bona fide desktop wiki which I want to store my technical research and howto docs (and share them in my digital garden).
This is what Obsidian is. If I wanted something to organize my tasks and other things, I'd take a much different route.
What I'm aiming to get is akin to these: [0] & [1]
It's a compiler problem and there is no money in compilers [1]. If someone made an intermediate representation for AI graphs and then wrote a compiler from that intermediate format into whatever backend was the deployment target then they might be able to charge money for support and bug fixes but that would be it. It's not the kind of business anyone wants to be in so there is no good intermediate format and compiler that is platform agnostic.
1. Reliable structured outputs 2. Reduced costs by 50% for input, 33% for output 3. Up to 16k output tokens compared to 4k
https://platform.openai.com/docs/models/gpt-4o