Show HN: A CUDA-Generating AI Agent

Hacker News - AI
Jul 9, 2025 23:02
atallahw
1 views
hackernewsaidiscussion

Summary

MakoGenerate is an AI agent that automatically converts PyTorch code into optimized CUDA kernels, iteratively improving performance and correcting errors based on feedback from compilation and validation. Users can customize the process with prompts, documentation, and specific optimization goals, making it a flexible tool for accelerating deep learning workflows. This approach has significant implications for automating and streamlining the development of high-performance AI models on GPU hardware.

MakoGenerate is an AI agent that takes any PyTorch code as input, and writes CUDA kernels for it. You can choose a sample problem or enter in your own PyTorch. It generates CUDA, compiles and validates the results, then uses all of this as feedback into the next generation attempt. If the generated kernel didn’t work, it tries to fix it. If it did work, it tries to make it faster. You can configure the agent with a model/hardware/compute budget. The optional custom prompt section gives you the opportunity to add more context to the agent. You can include specific documentation, tutorials, or examples. You can also identify specific optimization techniques that you would like to see applied. Prompts can be as short as one line, or even multiple pages long. Let me know what you think. Comments URL: https://news.ycombinator.com/item?id=44515598 Points: 2 # Comments: 0

Related Articles

PEPE Price Projection: $0.00005 by 2025 as Ozak AI Rockets to New Highs

Analytics InsightJul 17

The article discusses the projected rise of PEPE's price to $0.00005 by 2025, highlighting growing investor interest in AI-driven cryptocurrencies. It also notes that Ozak AI has reached new performance highs, signaling increased momentum and innovation in the AI sector. These trends suggest that AI integration is playing a significant role in shaping the future of digital assets and investment strategies.

Don't Fall for AI: Reasons for Writers to Reject Slop

Hacker News - AIJul 17

The article argues that writers should reject low-quality, AI-generated content ("slop") because it often lacks originality, depth, and emotional resonance. It warns that widespread use of such content could degrade creative standards and undermine the value of human authorship. This highlights ongoing concerns in the AI field about balancing technological advancement with the preservation of authentic, high-quality writing.

A £3.93/mo Nomad‑backed learning lab: Next.js · .NET · Postgres on a budget

Hacker News - AIJul 17

The article details how to set up a cost-effective learning lab for web development technologies like Next.js, .NET, and Postgres using Nomad, all for just £3.93 per month. While not directly focused on AI, this affordable infrastructure can support AI experimentation and learning by providing accessible backend resources for developers and students. The approach highlights the growing accessibility of robust development environments, which can accelerate AI prototyping and education.