BLOG ESPECIALIZADO NO MUNDO LINUX. Blog de Linux com vídeo aulas, notícias do mundo Linux e do Software Livre.
AMD ROCm 7.0.2 is now available, delivering critical stability patches & performance enhancements for AI/ML workloads and high-performance computing (HPC). This guide explores its new features, bug fixes, and impact on GPU-accelerated deep learning frameworks like PyTorch & TensorFlow
AMD’s new ROCm-LS toolkit and hipCIM library accelerate life science workloads on Instinct GPUs, offering CUDA compatibility & open-source alternatives for medical imaging, pathology, and AI-driven research. Explore early access now
AMD joins as a platinum sponsor for DebConf25, boosting ROCm support in Debian Linux. Discover how AMD’s open-source GPU computing strategy competes with NVIDIA CUDA, challenges in Debian’s packaging, and future AI/HPC developments.
AMD ROCm 6.4.1 is now available with official support for RDNA4 GPUs, including Radeon RX 9000 series. Discover new ROCm-DS for data science, installer updates, and compatibility with Ubuntu & RHEL.
AMD’s latest ROCm 7.0 preview enhances HIP Runtime for CUDA compatibility—key for AI developers using AMD Instinct GPUs. Learn how HIP updates improve ML workloads and GPU acceleration.
Discover AMD’s Radeon Test Drive—a cloud-based solution for developers to test ROCm on Radeon PRO GPUs before investing in hardware. Partnering with System76 and Colfax, AMD offers remote access to high-performance Threadripper workstations. Learn how this initiative compares to the AMD Developer Cloud and its potential impact on AI development.
AMD’s ROCm 2025 roadmap promises day-one Linux support for Radeon RX 9000 GPUs & Ryzen AI MAX. Discover how AMD plans to challenge NVIDIA CUDA with seamless AI/ML acceleration & enterprise-ready deployment.
O ROCm 5.7 foi lançado na sexta-feira, trazendo várias melhorias e novidades para a plataforma. Algumas das principais alterações incluem:
Fedora 42 update for AMD's ROCm RPP (rocm-rpp) removes the prebuilt libffts.a library, streamlining the high-performance computer vision package for HIP, OpenCL, and CPU backends on AMD processors. Learn the implications for developers and how to apply this update securely. | 178 chars
AMD merges True16 mode into LLVM for all RDNA3 GPUs (including RDNA 3.5), enabling native 16-bit registers & instructions. Boost ROCm compute & AI/ML performance potential. Explore the technical impact & future RDNA4 implications.
