Tag
AMD introduces the Instinct MI455X GPU, its first AI accelerator designed for rack-scale deployments, built on the new CDNA5 architecture with up to 40.26 PFLOP of performance, 432 GB of HBM4 memory, and the Helios rack-scale solution enabling 72-GPU pods.
A hands-on evaluation of the AMD Ryzen AI Halo, showcasing its AI acceleration capabilities and performance in real-world tasks.
OpenAI unveiled its first custom-built inference processor, named Jalapeño, developed with Broadcom to improve performance-per-watt and reduce reliance on Nvidia GPUs.
The LQ50 and LQ50-24GB are priced at around $1200, indicating a mid-range AI hardware offering.
The author open-sourced a custom AI accelerator (atik) implemented on FPGA with native BF16 and attention support, demonstrating significant speedups over PyTorch for various models.
PowerColor has released the Radeon AI PRO R9600D, a single-slot, passively cooled workstation GPU featuring 32GB of GDDR6 memory and a 12V-2x6 power connector.
WAN-IFRA and OpenAI launched the Newsroom AI Catalyst, a global accelerator program supporting 128 newsrooms across Europe, Asia Pacific, Latin America, and South Asia to adopt and implement AI technologies for improving content creation and operational efficiency.