Best of AMD2024

  1. 1
    Article
    Avatar of omgubomg! ubuntu!·2y

    Linux Kernel 6.9 Released, This is What’s New

    Linux kernel 6.9 has been released with new features such as AMD P-State 'preferred core' support, Intel Fastboot support, and an update to CPU timer code. It also includes performance improvements, bug fixes, and support for various devices.

  2. 2
    Article
    Avatar of jeffgeerlingJeff Geerling·2y

    LLMs accelerated with eGPU on a Raspberry Pi 5

    A stable patch for the `amdgpu` Linux kernel driver now allows AMD RX series GPUs (400 to 7000) to work with Raspberry Pi 5, supporting both Vulkan graphics and compute API. This guide provides steps to install `llama.cpp` with Vulkan support on the Pi 5. While larger models face performance issues due to inefficient memory access translations by the `amdgpu` driver, smaller models perform well. The compact, power-efficient setup only uses about 10-12W when idle, making it an attractive option for local machine learning tasks.

  3. 3
    Article
    Avatar of hnHacker News·2y

    HDMI Forum rejects AMD's HDMI 2.1 open-source driver

    AMD's efforts to integrate HDMI 2.1+ support into their open-source Linux graphics driver have been thwarted by the HDMI Forum's legal constraints. The rejection means Linux users with Radeon graphics cards won't be able to utilize advanced HDMI features like 4K@120Hz and 5K@240Hz. AMD recommends switching to DisplayPort for high-resolution display support.

  4. 4
    Video
    Avatar of techlinkedTechLinked·2y

    We’re In Trouble

  5. 5
    Video
    Avatar of techlinkedTechLinked·2y

    You’re gonna want this update.

    Tech news covers AMD's branch prediction improvements backported to Windows 11 version 23H2, offering significant gaming performance boosts for Ryzen CPUs. MSI motherboards now include a BIOS toggle to increase TDP for Ryzen 59600x and 7 9700x CPUs. The Google Pixel 9 Pro XL shows increased durability but is harder to repair. The FBI faces criticism over poor handling of electronic storage media. SpaceX delayed the Polaris Dawn Mission, and a Falcon 9 rocket fell into the ocean. AMD's new hardware leaks and Gannett shuts down its USA Today review site due to AI-written articles.

  6. 6
    Article
    Avatar of hnHacker News·2y

    Testing AMD’s Giant MI300X

    AMD's Radeon Instinct MI300X is designed to challenge NVIDIA's dominance in GPU computing. It leverages the massive CDNA 3 architecture, featuring 8 compute dies with extensive cache and memory setups. Compared to NVIDIA's H100, MI300X shows significant hardware advantages in cache configuration, bandwidth, and compute throughput. However, AMD's software ecosystem, specifically ROCm, is still lagging behind NVIDIA's CUDA, which is a critical factor for gaining widespread adoption. AMD aims to expand ROCm's compatibility to all its GPUs and CPUs to enhance its competitive edge.

  7. 7
    Article
    Avatar of hnHacker News·2y

    Stable Diffusion 3 Medium — Stability AI

    Stable Diffusion 3 Medium is Stability AI’s most advanced text-to-image open model yet, offering features such as photorealism, prompt adherence, typography, resource-efficiency, and fine-tuning. It is available for free trial and can be used for commercial purposes under certain licenses. Collaboration with NVIDIA and AMD has enhanced its performance.

  8. 8
    Article
    Avatar of jeffgeerlingJeff Geerling·2y

    AMD Radeon PRO W7700 running on Raspberry Pi

    The Pi community has successfully enabled AMD Radeon GPUs, including 6000 and 7000-series, to work on the Raspberry Pi 5. Some modern AAA games can run at low FPS on the Pi's hardware, while older games like Portal 2 run smoothly. This guide details how to set up and patch the Raspberry Pi OS for AMD GPU support, including firmware installation and kernel recompilation. It also addresses hardware transcoding support and provides a list of hardware used for the setup.

  9. 9
    Article
    Avatar of hnHacker News·2y

    Tune Llama3 405B on AMD MI300x (our journey)

    Felafax successfully fine-tuned the LLaMA 3.1 405B model on AMD MI300x GPUs using JAX, demonstrating impressive performance and scalability. By leveraging JAX's platform-independent optimizations and the device mesh feature for efficient parameter sharding, the team achieved near-linear scaling across 8 GPUs. This endeavor highlights AMD GPUs as a viable alternative to NVIDIA hardware for large-scale AI training, providing higher performance per dollar. The full open-sourced implementation is available on GitHub.