Archives de catégorie : Non classé

As Large Language Models continue to scale, the CPU host node in CPU-GPU AI accelerated systems has become an important performance component.

Enterprise AI on CPUs: Intel & T‑Systems prove RAG runs cost‑effectively without GPUs.

The landscape of AI agents is shifting from simple « ask-and-receive » interactions toward complex, multi-step reasoning.

When it comes to choosing the right foundation for your AI infrastructure, Intel Xeon 6 processors stand out as the clear winner.

Part 3 of our CPU:GPU blog series now moves from metaphor to operations.

This blog introduces a heterogeneous architecture that co-runs vLLMs on both CPUs and GPUs to improve overall system efficiency.

JAX-based Activation Offloading on Intel® Xeon® 6 with P-cores systems offers an effective alternative to activation recomputation, repurposing the CPU’s large DDR5 host memory as a live activation store and leveraging XLA’s asynchronous compute–communication overlap to avoid throughput loss.

By enabling priority cores to exceed standard all-core turbo limits and approach higher turbo frequency levels, PCT accelerates essential operations, reduces processing delays, and improves overall system responsiveness.

Confidential Computing is emerging as a foundational technology for Sovereign AI.

We explore the scalability and performance of agentic document summarization running on a Red Hat OpenShift cluster built on Lenovo ThinkSystem SR650 V4 servers powered by Intel Xeon 6 processors.