Friday, July 4, 2025

Agentforce London: Salesforce Reports 78% of UK Companies Embrace Agentic AI

WhatsApp Aims to Collaborate with Apple on Legal Challenge Against Home Office Encryption Directives

AI and the Creative Industries: A Misguided Decision by the UK Government

CityFibre Expands Business Ethernet Access Threefold

Fusion and AI: The Role of Private Sector Technology in Advancing ITER

Strengthening Retail: Strategies for UK Brands to Combat Cyber Breaches

Apple Encryption Debate: Should Law Enforcement Use Technical Capability Notices?

Sweden Receives Assistance in Strengthening Its Sovereign AI Capabilities

MPs to Explore Possibility of Government Digital Identity Program

CXL progresses as a memory technology for AI

At the recent Future of Memory and Storage summit in Santa Clara, Compute Express Link (CXL) was a focal point of discussion. CXL is an open standard technology designed for high-speed connections between processors and memory in datacentre servers. It utilizes the CXL.io protocol and PCIe 5.0 physical layer to provide low-latency interconnect paths, allowing for memory sharing and expansion without the need for additional infrastructure. CXL Memory Modules (CMMs) are particularly beneficial for AI applications that require high memory capacity, such as large language models.

Intel has noted how CXL can help lower hardware costs by allowing server manufacturers to use less expensive memory modules. Samsung has been collaborating with Red Hat to develop open source software for memory and storage products, including CXL memory, with a recent demonstration at the Red Hat Summit showcasing how CXL infrastructure can enhance performance in deep learning recommendation models. Samsung’s Scalable Memory Development Kit (SMDK) technology improves memory access performance and accelerates data processing and AI learning. Red Hat’s memory tiering function supports Samsung’s CMM-D memory modules, allowing for efficient allocation of memory based on usage patterns. This collaboration highlights the value of open source technology in advancing next-generation memory solutions.