Eduardo Bart

Emulators, compilers, Linux, C/C++, Lua, CG, DL, games and Web3. Developing RISC-V machine, and open source software.

[Eduardo Bart @edubart]

@morew4rd Also game developers, many game engines uses Lua as scripting, imagine for example you could require yobot Lua library in Lyte2D or Love2D, and then let your game NPCs truly have memory and interact with the game

[Eduardo Bart @edubart]

@morew4rd A new minimal/efficient/lightweight coding agent with core in C (or even C++) and plugins in Lua would be super cool. Everyday the desire that something like this should exist grows on me too, but I'm always too busy to play on the idea myself.

[Eduardo Bart @edubart]

AI just made me a presentation about Cartesi Machine (my daily work). Finally AI is handling the explanations so I can focus on the actual systems programming. Great overview if you're curious about what systems engineers like me are building. @cartesiproject @NotebookLM

[Eduardo Bart @edubart]

@GCdePaula_ @VitalikButerin Indeed! As someone who deeply works with RISC-V, I felt the need show some support there. I think engineers working with RISC-V should support the idea. More projects adopting RISC-V is a win for its ecosystem as a whole, paying back for everyone involved with RISC-V.

[Andrej Karpathy @karpathy]

"Move 37" is the word-of-day - it's when an AI, trained via the trial-and-error process of reinforcement learning, discovers actions that are new, surprising, and secretly brilliant even to expert humans. It is a magical, just slightly unnerving, emergent phenomenon only achievable by large-scale reinforcement learning. You can't get there by expert imitation. It's when AlphaGo played move 37 in Game 2 against Lee Sedol, a weird move that was estimated to only have 1 in 10,000 chance to be played by a human, but one that was creative and brilliant in retrospect, leading to a win in that game.

We've seen Move 37 in a closed, game-like environment like Go, but with the latest crop of "thinking" LLM models (e.g. OpenAI-o1, DeepSeek-R1, Gemini 2.0 Flash Thinking), we are seeing the first very early glimmers of things like it in open world domains. The models discover, in the process of trying to solve many diverse math/code/etc. problems, strategies that resemble the internal monologue of humans, which are very hard (/impossible) to directly program into the models. I call these "cognitive strategies" - things like approaching a problem from different angles, trying out different ideas, finding analogies, backtracking, re-examining, etc. Weird as it sounds, it's plausible that LLMs can discover better ways of thinking, of solving problems, of connecting ideas across disciplines, and do so in a way we will find surprising, puzzling, but creative and brilliant in retrospect. It could get plenty weirder too - it's plausible (even likely, if it's done well) that the optimization invents its own language that is inscrutable to us, but that is more efficient or effective at problem solving. The weirdness of reinforcement learning is in principle unbounded.

I don't think we've seen equivalents of Move 37 yet. I don't know what it will look like. I think we're still quite early and that there is a lot of work ahead, both engineering and research. But the technology feels on track to find them.

https://t.co/JCxTdKpuzv

[Eduardo Bart @edubart]

It embeds the Cartesi Machine emulator to run a full RISC-V Linux system. While networking is not yet supported, I have some ideas on how to support installing new software inside it in the future. Source code available at https://t.co/bA6SOMCfAB

[Eduardo Bart @edubart]

Project of the weekend, WebCM: A zero-install RISC-V Linux terminal running entirely in your browser! Built on Alpine Linux, it lets you experiment with multiple languages and tools in a sandbox. It runs completely client-side - no servers, no setup! https://t.co/DryfaJd5yQ

[Eduardo Bart @edubart]

@filpizlo Since both Fil-C and Rust use LLVM as backend, could Fil-C's safety features protect Rust's unsafe blocks? In other words, is Fil-C a general LLVM enhancement or is it C-specific?

[Eduardo Bart @edubart]

@filpizlo I think what you are trying to accomplish is amazing, it could be a big deal for the C/C++ ecosystem, looking forward to seeing how Fil-C evolves!

[Eduardo Bart @edubart]

@morew4rd @yacineMTB By the way, I updated sokol_gp.h to latest Sokol last week, along with some new additions people asked.

[Eduardo Bart @edubart]

@morew4rd Nice! I have been thinking lately we might have to start making gaming APIs and documentation with AI in mind, meaning we need to give more context and examples in its documentation so LLMs can perform better. Did it had any issues with Lua type system?

[Eduardo Bart @edubart]

@Max_Hatesuer @rives_io @dtuler @PaoloCTSI @nidaaBaris @lynoferraz More games on RIVES, yay! Eventually we will have a large library of games!

[Eduardo Bart @edubart]

@FlohOfWoe Watching the animation of the chip circuit working in real time is so cool

[Eduardo Bart @edubart]

@FlohOfWoe Nice! My only wish is that you went further and turned this in a single header C++ library, or a two file library (cimgui.cpp + cimgui.h) so is more minimal to download and use (in style of Sokol headers).

[Eduardo Bart @edubart]

@Max_Hatesuer @rives_io Nicee, I am surprised how far you and others got with no code experience, by using just IA and simple tools.

[Gabriel Coutinho de Paula 🐝 🐧 @GCdePaula_]

Took us a bit longer than expected, but Dave is finally here!

We published a novel fraud-proof algorithm, that is truly decentralized while being resistant to Sybil attacks:

https://t.co/3NVBwTktfk

We’ll present our findings @EFDevcon! Come meet us Wednesday 11h at stage 5.

[RIVES 🟪 @rives_io]

We brought DOOM (actually) onchain. Now, get ready for the DOOM Olympics, live on @base Mainnet!

🗓️ Sep 12-19
🏆 $15k prize pool sponsored by @cartesiproject
🤼 7 contests using the Freedoom build
✅ Verifiable gameplays and scores

Sign Up: https://t.co/iBf5FOXENK
More info 👇

[Eduardo Bart @edubart]

@ednekebno Well done! I Had fun playing it! Impressive work within the amount of time and constrains!

[Eduardo Bart @edubart]

@ednekebno Ideas to optimize:
1. Make sure to compile with CFLAGS from `riv-opt-flags -Ospeed`
2. Set `riv->draw.color_key_disabled = true` when drawing fully opaque sprites (e.g background).
3. Drop the frame rate from 60fps to 30fps.
4. Last resort, lower the game resolution

[Eduardo Bart @edubart]

@ednekebno Very cool! Vampire survivors-like for RIVES! Keep it going! I am surprised by how much you did there already!