Intel just announced the 4th generation Intel Xeon Scalable processors and Intel Xeon CPU Max series, which are coming to market today. They incorporate the latest Intel accelerator engines for AI, high-performance computing (HPC), security, networking, analytics, and storage. Integrated acceleration is presented as a more efficient alternative to achieve higher performance than increasing the number of CPU cores.
The latest Intel Accelerator Engines, high-bandwidth memory, and software optimizations help improve performance and power efficiency in specific workloads. They can also generate cost savings through better CPU utilization.
Table of Contents
Accelerators built into 4th Gen Intel Xeon
Intel® Advanced Matrix Extensions (Intel® AMX)
The new Intel AMX extends AI performance in CPUs to include fine tuning and small and medium deep learning training models. This built-in accelerator improves deep learning inference and training performance. It is presented as an ideal accelerator for workloads such as natural language processing, recommender systems and image recognition.
Intel® QuickAssist Technology (Intel QAT)
By offloading encryption, decryption, and compression, Intel QAT, which is now integrated as a built-in accelerator; helps free up the processor core so that the systems can serve a greater number of customers or that they can consume less energy. With Intel QAT, 4th generation Intel Xeon Scalable processors are the highest performing CPUs that can compress and decrypt data in a single data stream.
Intel Data Stream Accelerator (Intel DSA)
Intel DSA drives high performance for storage, networking, and data-intensive workloads by improving streaming data movement and transformation operations. It is designed to offload the most common data movement tasks that cause overhead in data center-scale deployments. Thus, Intel DSA helps speed the movement of data across the CPU, memory, and caches, as well as across all connected memory, storage, and network devices.
Intel Dynamic Load Balancer (Intel DLB)
Intel DLB helps improve system performance related to network data handling on multicore Intel Xeon Scalable processors. This allows Efficient distribution of network processing across multiple CPU cores and threads and a dynamic distribution for processing as system load varies. It also restores the order of the network data being processed simultaneously across the CPU cores.
Intel In-Memory Analytics Accelerator (Intel IAA)
Intel IAA helps run database and analytics workloads faster. And it does so with potentially greater energy efficiency. This built-in accelerator increases the performance of Big Data queries and reduces memory consumption for in-memory databases. Intel IAA is ideal for in-memory databases, open source databases, and data warehouses like RocksDB and ClickHouse.
Intel 512 Advanced Vector Extensions (Intel AVX-512)
Intel AVX-512 is the latest x86 vector instruction set, with up to two Fused Addition and Multiplication (FMA) units and other optimizations to accelerate performance for demanding computing tasks such as scientific simulations, financial analysis, and modeling and 3D analysis.
Intel® Advanced Vector Extensions 512 (Intel® AVX-512) for vRAN
Intel AVX-512 for Virtualized Radio Access Networks (vRAN) is designed to offer higher capacity at the same power consumption for vRAN workloads. This helps communications service providers to increase your performance per watt to meet critical performance, scaling, and energy efficiency requirements.
Intel Crypto Acceleration
Intel Crypto Acceleration reduces the impact of deploying pervasive data encryption while increasing performance for encryption-sensitive workloads. Some of the use cases are Secure Sockets Layer (SSL) web servers, 5G infrastructure, and VPNs and firewalls.
Intel Speed Select Technology (Intel SST)
Intel SST is designed to give you more active and expansive control over CPU performance. It improves server utilization and reduces qualification costs by allowing customers to configure a single server to accommodate fluctuating workloads. Thus a flexible server with multiple configurations is achieved.
Intel® Direct Data I/O (Intel® DDIO) Technology
Intel DDIO helps eliminate inefficiencies by enabling direct communication between Ethernet adapters and controllers and the host processor cache. Eliminating frequent visits to main memory can help reduce power consumption, as well as provide greater I/O bandwidth scalability and lower latency.
Intel Security Engines in 4th Gen Xeons
Intel Software Guard Extensions (Intel SGX)
With Intel SGX, you can open up new opportunities for business insight and collaboration, even with sensitive or regulated information. Intel SGX is the most researched, updated, and deployed confidential computing technology in data centers on the market today, with the smallest trust boundary. confidential computing improves isolation of sensitive data to
Enhanced hardware-based memory protections.
Intel Trusted Domain Extension (Intel TDX)
Intel TDX is a new capability available through select cloud providers in 2023 that offers greater confidentiality at the virtual machine level. This improves privacy and control over data. Within an Intel TDX Confidential VM, the VM's guest OS and applications are isolated from access by the hypervisor and other virtual machines on the platform.
Intel Control-Flow Enforcement Technology (Intel CET)
This provides enhanced hardware-based protections against callback and jump/call-oriented programming attacks, which are among the most common software-based attack techniques. Using Intel CET helps end a whole class of system memory attacks that have long eluded software-only solutions.
Intel also announces On Demand
What is Intel OnDemand?
Intel® On Demand is a new service being introduced with XNUMXth generation Intel® Xeon® Scalable processors. On Demand can be used to upgrade and/or upgrade hardware enhancements and accelerators available on most XNUMXth generation Xeon processor SKUs.
Supported features include Intel® Dynamic Load Balancer, Intel Data, Intel Data Streaming Accelerator, Intel In-Memory Analytics Accelerator, Intel® Quick Assist Technology, and Intel Software Guard Extensions.
How does Intel On Demand work?
It consists of an API for license ordering and a software agent for license provisioning and CPU feature activation. Depending on the hardware vendor, the end customer will have the option to choose to purchase On Demand features at the time of purchase or after purchase such as during the upgrade. Intel has referred to this adoption model as “activation.”
Alternatively, Intel is also working with some partners to implement a metered adoption model where on-demand features can be turned on and off as needed so that payment is based on usage rather than a one-time license. Intel has referred to this adoption model as “consumption.”
What are the supposed benefits of On Demand for customers?
Intel has heard from end customers who want to move from capex to opex. Therefore, it will align better with demand and budget constraints. end customers want access to new features and capabilities when they need them. With the introduction of the activation model with the processors
4th generation Xeon processors, On Demand will offer end-customers the flexibility to choose premium full-featured SKUs or add only those needed at any time throughout the Xeon processor lifecycle.
Intel believes that the flexible nature of On Demand will become more valuable to customers as the number of workloads and the amount of third-party enterprise software using accelerators increases.
