
The 4th Gen Xeon Scalable CPU from Intel (NASDAQ: INTC) Lays Forth a Plan For Universal AI
CIO Review APAC | Thursday, January 19, 2023

Intel (NASDAQ: INTC) is plotting a path to bridge that divide with its 4th Gen Xeon Scalable CPUs.
FREMONT, CA:Machine learning (ML) and artificial intelligence (AI) applications have traditionally had different workloads that may operate on CPUs and GPUs. With its 4th Gen Xeon Scalable CPUs, Intel (NASDAQ: INTC) is laying out a plan to cross that gap.
ML training has always been viewed as the sole responsibility of GPUs and other hardware designed specifically as an accelerator. That is the predicament Intel is currently trying to change. The objective of Intel is to make it possible for businesses of all sizes to employ CPUs for training, AI inference, and data preparation on a single data centre platform that works the same from the network edge to the cloud.
Stay ahead of the industry with exclusive feature stories on the top companies, expert insights and the latest news delivered straight to your inbox. Subscribe today.
AI is now permeating every application and workflow. By concentrating on the end-to-end performance of the application, we hope to speed this AI infusion into every application.
Intel is introducing its Sapphire Rapids-codenamed 4th generation Xeon Scalable CPUs today to achieve that objective. Numerous new features built into the new processor are intended to speed up AI tasks on Intel's CPU. The launch of Intel's AI Software Suite, which offers both open-source and paid tools to support the development, deployment, and optimization of AI workloads, coincides with the release of the latest silicon update.
Intel Advanced Matrix Extensions (AMX) Accelerates AI
The inclusion of Intel Advanced Matrix Extensions is one of the 4th generation Xeon Scalable processor's key AI breakthroughs (AMX). The core of many deep learning workloads nowadays is dense matrix multiplication, which is accelerated by the CPU via Intel AMX.
Many firms now use discrete GPUs to offload their inferencing requirements to achieve the appropriate degree of performance and service level agreements. When compared to Intel third-generation Xeon processors, the Intel AMX can deliver a 10x performance gain in AI inference speeds. Training and data preparation are both sped significantly by the new Intel CPU.
The Path to Transfer Learning
In recent months, generative AI and large language models (LLMs) have received a lot of attention in the AI community.
The initial training for LLMs will often still require a discrete GPU, such as those from the Intel Max Series, claims Intel. However, the Intel AMX capabilities will offer great performance for more typical use situations where a company wants to improve an existing LLM or retrain an existing model. Additionally, Intel is promoting transfer learning as a key use for Intel AMX in this field.
Modern AI workloads cannot be enabled by hardware alone. Also required is software.
The new Intel AI Software Suite, which combines open-source frameworks, tools, and services to assist enterprises in building, training, deploying, and optimising AI workloads, is how Intel is currently coordinating its AI software activities.
The Intel Developer Catalog is one of the technologies included in the AI Software Suite. The senior director of Intel AI products revealed that the catalogue offers 55 pre-trained deep learning models that users can immediately download and use in a press briefing.
The package also contains SigOpt, a piece of technology that Intel bought in October 2020. During the ML training phase, SigOpt offers tools for hyper-parameter adjustment. The Intel AI Software Suite also includes OpenVINO, which supports businesses in developing and implementing models.
Intel is certain that its 4th Gen Xeon Scalable CPU will assist to democratise AI by making it more broadly available and usable to the mix of software and hardware that is easily deployed in the data centre, cloud, and edge locations.
More in News