Gain Insights into Intel® Xeon® 6 with P-Cores: Built for AI and Analytic Workloads
The Intel® Xeon® 6700-series processor with P-cores is here to round out the complete Intel® Xeon® 6 with P-cores offering. There has never been a greater demand on data center performance, from compute-intensive AI and HPC use cases to mission-critical relational database and analytics applications. Intel® Xeon® 6 processors are built to empower IT teams with targeted enhancements for a variety of workloads across two product line offerings: P-core and Efficiency-core (E-core) offerings.
P-core offerings are differentiated with hyperthreaded cores featuring built-in matrix engines that accelerate compute-intensive AI, HPC, and data services workloads, whereas E-core offerings omit matrix engines in favor of higher, single-threaded core density.
All Intel® Xeon® 6 processors, regardless of P-core or E-core focus, feature the same instruction sets, BIOS, and built-in I/O accelerators—including Intel® QuickAssist Technology (Intel® QAT), Intel® Data Streaming Accelerator (Intel® DSA), Intel® In-Memory Analytics Accelerator (Intel® IAA), and Intel® Dynamic Load Balancer (Intel® DLB).
They also support the same drivers, operating systems, and applications and the same reliability, availability, and serviceability (RAS) capabilities that data centers rely on.
Explore Intel® Xeon® 6 Processor with P-Core Advantages
Businesses are looking to save money and time in deploying state-of-the-art AI models for language, vision and image recognition, and recommendation systems, as well as to generate deeper insights from HPC-fueled simulation and advanced analytics. Intel® Xeon® 6 processors with P-cores empower businesses to act on these opportunities by providing a large number of high-performance cores with innovative matrix and vector engines as well as ultrahigh memory and I/O bandwidth.
Deploy High Performance for a Vast Range of Workloads
With the flexibility to scale up to 128 cores, 12 memory channels, and 96 PCIe lanes per socket, Intel® Xeon® 6 processors with P-cores help businesses meet their varying application requirements, on prem or in the cloud. For IT teams looking to alleviate memory bandwidth bottlenecks, innovative Multiplexed Rank DIMM (MRDIMM) delivers up to 8,800 mega transfers per second (MT/s) while contributing to low TCO by completing jobs fast. Built-in accelerators give an additional boost to targeted workloads for even greater performance and efficiency.
Intel® Xeon® 6700/6500-series with P-cores is built for scale, offering 4S/8S options as well as a 1-socket SKU set. The 1-socket platforms deliver significantly more I/O in a single-socket, via an increase in PCIe lanes for peripherals and storage devices - 136 compared to 88 lanes per socket in a multi-socket platform. This improved I/O per socket helps to remove power/latency inefficiencies, and improves platform cost optimization. Customers can improve TCO if their I/O requirements can be met with a single-socket platform – fewer required CPUs leads to reduces server cost, improved performance/watt and less per-CPU licensing costs.
Take Advantage of Powerful AI Performance from a CPU
Intel® Xeon® 6 processors with P-cores are designed to support many demanding AI use cases. P-cores expand on four generations of Intel’s leadership in built-in AI with acceleration such as Intel® Advanced Matrix Extensions (Intel® AMX), which now support int8, BF16, and FP16 (new) data types. As a result, P-cores help to meet service-level agreements (SLAs) for several AI models, ranging from object detection to midsize GenAI, while offering open standards, high performance, RAS features, and support for additional accelerators as needed.
Intel® Xeon® 6 with P-cores also serves as the best host CPU option paired with AI accelerators in AI accelerated systems. Host CPUs are expected to deliver superior I/O performance, high single threaded performance, and high memory bandwidth & capacity. Learn more about the top 5 reasons why Intel® Xeon® 6 with P-cores is the best host CPU option.
Optimize Performance-per-Watt for General Purpose Workloads
Intel® Xeon® 6 processors with P-cores excel at the complete spectrum of workloads, with a mainstream series that features a range of eight to 86 cores in the mainstream offering, up to 176 PCIe 5.0 lanes for networking and storage add-in cards in dual CPU-based systems, and a single-socket offering with a remarkable 136 PCIe lanes for single CPU-based systems. The efficiency of all Intel® Xeon® 6 processors is highlighted by their ability to provide scalable performance per watt as server utilization increases, delivering nearly linear power-performance consumption across the load line. For performance-demanding workloads, this means the platform efficiently uses power at high loads to help finish jobs fast.
Performance-core Workload Performance Benefits:
Compared to 5th Gen Intel® Xeon® Processor:
- 40% average performance gain across broad workloads1
- Up to 3x better Llama2 performance with Intel® AMX2
- Up to 2x better HammerDB MySQL performance3
- Up to 1.58x higher Performance per Watt at a typical 40% server utilization point4
- Up to 1.45x higher database analytics performance5
- 1.5x better performance and performance/watt for NGINX edge workloads6
- Up to 2.5x better performance for HPCG benchmark with MRDIMM7
Compared to 2nd Gen Intel® Xeon® Processor:
- Up to 17x better performance for ResNet – 50 BSX with INT88
- 3.9x average general compute performance1
- 17:1 server consolidation Intel Xeon 6900 series compared to 2nd Gen Intel Xeon at maintained performance level9
- 5:1 average server consolidation Intel Xeon 6700 series compared to 2nd Gen Intel Xeon at maintained performance level with an average TCO gain of 40%10
Compared to Competition:
- Up to 2x average AI performance gains vs AMD EPYC 975511
- Up to 1.52x higher HPCG performance vs. AMD EPYC 965512
Keep Up with Business Growth with Enhanced Security Features
Pursue new business models and data sharing across on-premises, edge, and cloud servers, even when working with sensitive or regulated data. Confidential computing with trusted execution environments (TEEs) helps protect data and AI models while in use. Intel® Xeon® 6 processors with P-cores allow customers to pick the confidential computing technologies that best meet their business and regulatory requirements.
Both Intel® Xeon® 6 with P-cores products support Intel® TDX Connect. Intel® Trust Domain Extensions (Intel® TDX) is our confidential computing technology that protects workloads at the virtual machine (VM) level. First disclosed at the Open Confidential Computing Conference (OC3) in early 2023, Intel TDX Connect enables seamless encrypted communications between a Confidential VM and an enabled PCI Express device. This capability is critical to enabling confidential usage models that extend beyond the CPU to connected devices such as GPUs, Smart NICs or storage drives. Intel TDX Connect is a simple extension of the architecture designed to enhance confidential computing infrastructure, and is also designed to improve I/O virtualization performance.
Intel is rapidly collaborating with partners to expand the deployment of Intel TDX Connect capabilities throughout the partner ecosystem. Microsoft has already committed to enabling Intel TDX Connect support in Microsoft Azure instances.
"Microsoft is excited to productize Intel TDX Connect into future generations of Azure confidential VMs, which is in early development between our hardware and software developers. Intel TDX Connect represents a major milestone in our journey to improve confidential computing performance and extensibility. It enables existing and net-new workloads to benefit from the privacy assurances of confidential computing without compromising on price or performance."
-Vikas Bhatia, Head of Product for Azure Confidential Computing at Microsoft
Additionally, Intel is actively collaborating with other device & accelerator vendors to ensure they support PCIe protocols that underpin Intel TDX Connect, allowing for the successful expansion of Intel TDX-based confidential computing capabilities for a variety of customer needs.
- Application isolation: Intel® Software Guard Extensions (Intel® SGX) provides application isolation designed to protect data in use. Intel® SGX is the most researched and updated confidential computing technology in data centers on the market today.
- Virtual machine (VM)‒level isolation: Intel® Trust Domain Extensions (Intel® TDX) offers isolation and confidentiality at the VM level. Within an Intel® TDX confidential VM, the guest OS and VM applications are isolated from access by the cloud host, hypervisor, and other VMs on the platform.
Enhance Targeted Workloads Optimized for P-Cores
IT decision-makers who manage compute-intensive workloads for use cases such as AI, HPC, and relational databases will see the most impact from deploying Intel® Xeon® 6 processors with P-cores. Added support for MRDIMM memory can also help alleviate memory access constraints for use cases such as in-memory databases or large language models (LLMs) with outsized memory footprints.
Accelerate AI Workflows to Stay Competitive
Together with enhanced cores, higher memory bandwidth, and powerful matrix engines, Intel® Xeon® 6 processors with P-cores deliver ample compute to support small to midsize generative AI models for inferencing, fine-tuning, and retrieval-augmented generation (RAG) use cases. Additionally, optimizations for Intel® Xeon® processors are already integrated into the mainstream distributions of popular deep learning frameworks, including TensorFlow and PyTorch.
Manage Increasingly Complex HPC Calculations and Simulations
In addition to built-in matrix engines, Intel® Xeon® 6 processors with P-cores feature Intel® Advanced Vector Extensions 512 (Intel® AVX-512) to accelerate vector-based mathematics common to scientific simulations. Clusters powered by this platform can help unleash the power of 3D rendering, scientific research, financial simulations, and computer-aided engineering to speed time to results.
Achieve New Levels of Infrastructure and Storage Optimization
Because P-cores and E-cores share the same hardware platform, organizations can mix systems with each type of core to manage a diverse range of workloads in their infrastructure while staying within a power budget and adhering to data privacy and sovereignty regulations. Compute-intensive P-cores can support intense business analysis software such as SAS, while E-cores can be used to run low-intensity, high-idle-time workloads such as system backups and software updates.
Boost Productivity and Help Eliminate Latency at the Edge
Enhance edge productivity and latency by placing high performance compute resources closer to end users and devices. Intel® Xeon® 6 processors with P-cores enable the deployment of advanced AI models in edge servers and provide the compute power needed to deliver fast insights from thousands of IoT devices. Single-socket options with 136 PCIe lanes also help consolidate infrastructure in space-constrained deployments by allowing for more add-in cards per single processor.
Grow Relational Databases while Maintaining Fast, Responsive Analytics
Advanced vector engines in P-core processors allow single instruction, multiple data (SIMD)-biased workloads common to advanced database and analytics use cases to run effectively. Intel® Xeon® 6 processors also take advantage of a high level of exceptionally low‒latency cache to help avoid delays associated with resource contention. Versatile configurations with this platform and accelerators such as Intel® QAT can produce fast results when analyzing and storing complex data types.
Learn More About Intel® Xeon® 6 Processors
There are many advantages to deploying both Intel® Xeon® 6 processors with E-cores and P-cores. To better understand the technical and competitive advantages of both, download and read this analysis by Prowess Consulting, and contact your Intel representative to get started.