Hacker News·4yan open-source GitHub Copilot server
FauxPilot is an attempt to build a locally hosted version of GitHub Copilot. It uses the SalesForce CodeGen models inside of NVIDIA's Triton Inference Server with the FasterTransformer backend. If you have two NVIDIA RTX 3080 GPUs, you should be able to run the 6B model by putting fauxpilot-triton-1. The container image is governed by the NVIDIA Deep Learning Container License. Don't use it.
omg! ubuntu!·4yNVIDIA Unexpectedly Announces Open-Source GPU Kernel Modules
NVIDIA announces the immediate release of open source GPU kernel modules for a crop of its current hardware. The move, it says, is the first step in a broader open source by the company aimed at ‘improving the experience of using NVIDIA GPUs in Linux’ NVIDIA trails its rivals (mainly AMD and Intel, though bonus points if you remember VIA) in support for emerging open source technologies.
The New Stack·4yBrain.js Brings Deep Learning to the Browser and Node.js
Brain.js is a JavaScript library for deploying a neural network in the browser or on Node.js. It uses a computer’s GPU (graphics processing unit) for calculations, or pure JavaScript when GPU isn’t available. A lot of deep learning projects will continue to use the likes of TensorFlow or big company platforms like IBM Watson Studio or Google Cloud AI Platform.
Pointer·4yFree and Open Source AI Image Upscaler for Linux, MacOS and Windows built with Linux-First philosophy.
Free and Open Source AI Image Upscaler for Linux, MacOS and Windows built with Linux-First philosophy. Upscayl uses AI models to enhance your images by guessing what the details could be. It uses Real-ESRGAN (and more in the future) model to achieve this.
Hacker News·4yGPU Mining No Longer Profitable After Ethereum Merge
The best graphics cards should finally be back where they belonged, in your gaming PC. That's a quick drop, considering yesterday there were still a few cryptocurrencies that were technically profitable. Buying a used graphics card looks like it's set to become far. This often means keeping the cards near their limits by using fan curves that keep them near their thermal limits.
Hacker News·3yTales of the M1 GPU
Alyssa Rosenzweig started working on reverse engineering the M1 GPU in April. She started writing an M1 driver for Mesa, the Linux userspace graphics stack, and just a few months later she was already passing 75% of the GL2 conformance 2 conformance. Earlier this year, her work was so far ahead.
Hacker News·4yApple unveils M2 with breakthrough performance and capabilities
Apple unveils M2, taking the breakthrough performance and capabilities of M1 even further. M2 has an 18 percent faster CPU, a 35 percent more powerful GPU, and a 40 percent faster Neural Engine. It also delivers 50 percent more memory bandwidth compared to M1 and up to 24GB of fast unified memory.
Hacker News·3yTen Minute Physics
In this tutorial I explain the basics of Eulerian, grid-based fluid simulation and show how to write a simulation engine based on these discussed concepts. I use python in connection with the n Nvidia python extension warp as a simple way to write GPU simulations. I also briefly discuss the implementation as a python blender plugin.
80 LEVEL·4yNVIDIA Maxine Maintains Eye Contact for You
NVIDIA Maxine Maintains Eye Contact for You NVIDIA announced the re-architecture of Maxine for cloud-native microservices. Maxine’s cloud- native microservices allow developers to build real-time AI applications. The service enables clear communications to enhance virtual interactions.