World Tech

 By Nadim Kahwaji Matrix operations, particularly multiplication, are the backbone of the artificial intelligence (AI) world, playing a crucial role in today's AI models. These operations underpin the technologies behind speech and image recognition, chatbots from every major vendor, AI-driven image generators, and advanced video synthesis models. Beyond AI, matrix math is fundamental to modern computing applications such as image processing and data compression. Given the compute-intensive nature of matrix multiplication, it places significant demands on hardware systems, whether deployed in sprawling data centers or localized on-premises machines or your mobile. In the context of training AI models or generating images, it is not uncommon for the process to involve millions of matrix multiplication operations. This staggering number highlights the intensive computational workload required, where any advancement promising to streamline matrix multiplication operations could have a...

 By Nadim Kahwaji Intel's latest processors, Granite Rapids-D and Sierra Forest, unveiled at MWC 2024, are well-suited to meet the demands of Virtualized Radio Access Networks (vRAN) and offer substantial benefits to Software-Defined Networking (SDN) environments. vRAN is an approach that deploys radio access network functions on general-purpose hardware instead of specialized equipment. This strategy enables mobile network operators to scale and modify their infrastructure in response to evolving demands, especially for today’s 5G workloads, potentially leading to significant cost savings flexibly and efficiently. SDN, on the other hand, separates the network's control logic from the physical routers and switches through which data packets flow. The architecture of SDN features a centralized controller, typically installed on servers, which directs network traffic and makes decisions about how data should flow through the network. This setup is hardware-agnostic, but the n...