Table of Contents 64q43
The new generation of Intel Xeon was already announced a few months ago in the Intel Innovation and, this Thursday (14), Pat Gelsinger, CEO of the company, presented more details about the brand new processor during the AI Everywhere, an event focused on AI and how Intel intends to accelerate the arrival of this new world. O 5th Generation Intel Xeon promises faster memory and performance improvements with a focus on keeping up with the pace of evolution of artificial intelligence, preparing data centers to receive and process a huge amount of data, giving way to a new era of global expansion in which computing power is even more important for everyone's better future. 1q3t2u
How important is the Data Center for AI? 1y3265
Before we see the new impacts that the 5th Generation Intel Xeon can bring to data centers, it is important to understand the relevance of these places in driving the continuous evolution of artificial intelligence (AI). These server complexes, often distributed globally, play a vital role in storing, processing, and managing the enormous volumes of data needed to train and improve AI algorithms.
The importance of data centers for AI is vital. Firstly, they provide the infrastructure needed to host vast amounts of data, from training sets to real-time data. Storage capacity and efficiency in data retrieval are crucial aspects to ensure that artificial intelligence models have quick and efficient access to the information needed to make accurate decisions.
Furthermore, data centers are extremely important for training complex AI models. Modern algorithms such as deep neural networks require enormous amounts of computing power to tune and optimize millions or even billions of parameters. Data centers, often equipped with high-performance graphics processing units (GPUs), make it possible to run these intensive calculations in parallel, significantly speeding up the training process.
Currently, the most important use of data centers in the world of AI is its implementation through cloud computing – or cloud computing -, making it possible to process this data at absurd speeds, without the need for the or company to have their own data server. We will see below how companies like Google have implemented the new chip Intel to improve your cloud services.
As processor manufacturers such as Intel, develop and improve their architectures, the efficiency and processing power of the chips increase. This not only allows for faster execution of AI tasks, but also enables the development of more sophisticated and complex models. The improved processing capacity of 5th Generation Intel Xeon is fundamental to dealing with the increasing computational demands of artificial intelligence.
The impact of 5th Gen Intel Xeon 615v1y
O Intel Xeon 5th Generation represents a significant leap in performance and efficiency for data centers, clouds, networks and edge environments. This new processor offers an average performance gain of 21% while maintaining the same thermal power, resulting in an impressive 36% increase in performance per watt. For customers following a typical five-year upgrade cycle and transitioning from older generations, the potential reduction in Total Cost of Ownership (TCO) reaches up to 77%.
Built for AI, 5th Gen Intel Xeon processors deliver greater performance for customers deploying AI capabilities in the cloud, network, or edge devices. As a result of our long-standing work with customers, partners and the developer ecosystem, we are launching the 5th Gen Intel Xeon on a proven solid foundation that will enable rapid adoption and scale with lower TCO.
Sandra Rivera, Intel executive vice president and general manager of the Data Centers and AI Group
One of the highlights of 5th generation Intel Xeon is its unique integration of Artificial Intelligence (AI) acceleration directly into the processor. This translates into up to 42% more inference and fine-tuning capacity on models with up to 20 billion parameters. Also, the Xeon is the only U with a consistent and constantly improving set of performance results benchmarking training and inference MLPerf, ensuring reliability and efficiency in Machine Learning.
AI accelerators built into Xeon, combined with optimized software and advanced telemetry capabilities, facilitate manageable and efficient deployments of demanding workloads across diverse industries. This includes communications service providers, content delivery networks and vertical markets such as retail, healthcare and manufacturing.
During launch, the IBM highlighted that processors 5th generation Intel Xeon achieved up to 2,7x better query performance on their platform watsonx.data, compared to processors Xeon of the previous generation. O Google Cloud, which plans to implement the 5th generation Xeon next year, reported that the Palo Alto Networks doubled the performance of its deep learning models for threat detection using built-in acceleration 4th generation Xeon by means of Google Cloud.
Additionally, the game studio Gallium Studios leveraged the company’s AI platform numenta, which uses processors Xeon, achieving a 6,5x increase in inference performance on a GPU-based cloud instance, resulting in reduced costs and latency in your game proxi AI-based.
Based on previous generations, the new processor brings significant innovations in of performance and efficiency. These processors have the ability to operate with up to 64 cores per U, representing almost triple the maximum last-level cache compared to the previous generation. Additionally, they offer eight channels of DDR5 per U, ing DDR5 with transfer rates of up to 5.600 megatransfers per second (MT/s), and expand bandwidth between sockets with Intel UPI 2.0, providing up to 20 gigatransfers per second (GT /s).
In addition to improved performance, processors 5th generation Intel Xeon are recognized as the most sustainability-focused data center processors ever developed. With built-in accelerators that contribute to improved performance per watt, energy efficiency and savings can be further extended by enabling Power Optimized Mode in the platform BIOS, especially in workloads that can benefit from this feature.
In of security, the Intel Trusted Domain Extensions (Intel TDX) offer a higher level of confidentiality in the virtual machine (VM), improving privacy and providing greater control over data. On the other hand, the Intel Software Guard Extensions (intel sgx) represent the leading technology in confidential computing, being the most researched, updated and comprehensive available in data centers on the current market.
Processors 5th generation Xeon are backward compatible with previous generation pins, and important original equipment manufacturers (OEMs) such as Cisco, Dell, HPE, IEIT Systems, Lenovo, Super Micro Computer, among others, plan to offer a wide range of processors for single- and dual-socket systems starting in the first quarter of 2024. Additionally, leading Cloud Service Providers (CSPs) will announce the availability of instances based on 5th generation Intel Xeon throughout the year.
This advance in the performance of 5th generation Intel Xeon opens up new possibilities for natural language processing and advanced AI applications, not just in data centers and clouds, but also across network applications and at the edges, driving innovation and efficiency across multiple industries.
More demand for Generative AI with Intel Gaudi3 4b3v6a
In addition to the new chips, the Intel also provided more details about the Intel Gaudi3, new version of the company's AI acceleration processor. The company plans to launch the accelerators in 2024 and this new hardware will be built on the high-performance architecture of gaudi2, promising to deliver 4x more computing power (BF16), double the network bandwidth for more effective scaling, and 1,5x more integrated HBM memory. These improvements aim to meet the growing demand for high-performance computing and efficiency of LLMs, without compromising performance.
During the AI Everywhere, Gelsinger first presented an accelerator wafer gaudi3 and announced that the Intel is experiencing a significant expansion in its research and development of the Gaudi due to growing and proven performance advantages, combined with highly competitive prices. Given the growing demand for Generative AI solutions, Intel plans to capture an even larger share of the accelerator market in 2024, driven by its suite of AI accelerators, highlighted by Intel Gaudi3.
Watch the video on the Showmetech channel:
Source: Intel
reviewed by Glaucon Vital in 14 / 12 / 23.
