Every year the demand for higher-performing processors grows, whether for data servers, laptops, or desktop computers. These new processors must deliver greater functionality with more CPU cores, lower latency, higher performance per workload, and improved security.
The two most common processor architectures are the x86 and Arm. Intel originally introduced x86, which is mainly based on complex instruction set computing (CISC) architecture with backward compatibility, while Arm is based on reduced instruction set computing (RISC) architecture.
CISC architecture (left) and RISC architecture (right). Image used courtesy of JavaTpoint
Most recently, Intel has introduced its 3rd generation Xeon Scalable processors with built-in AI for data center applications.
Intel’s Latest Processor for Data Center Platforms
Intel’s new processor launched last week. The company claims this device is uniquely optimized for the requirements of hybrid-cloud, high-performance computing, networking, and intelligent edge applications.
This x86 data center processor (also called “Ice Lake“) features built-in AI acceleration, data protection platform (Intel SGX), and crypto acceleration.
Intel’s 3rd Xeon x86 data center processor. Image used courtesy of Intel
Based on 10nm process technology, it comes with up to 40 cores per processor. This platform also supports up to 6 TB of system memory, eight channels of DDR4-3200 memory, and 64 lanes of PCIe Gen4 per socket.
How Does Xion Stack Up Against AMD’s EPYC Processor?
Last month, AMD released its 3rd Gen EPYC 7003 Series CPUs built on the “Zen 3” core and AMD Infinity Architecture using 7nm process technology.
This x86-based processor is targeted at data centers serving hyper-converged infrastructure (HCI), database, and big data analytics, as well as high-performance computing workloads.
Compared to Intel’s 3rd Gen Xeon processor, which comes with up to 40 cores per processor and a base clock of 2.3 GHz, AMD’s latest processor features up to 64 cores with a base clock of 2.45 GHz.
AMD’s EPYC 7003 architecture. Screenshot used courtesy of AMD
According to Intel, the 3rd Gen Xeon Scalable processor delivers 1.5 times higher performance across 20 popular AI workloads compared to AMD’s EPYC 7763. The EPYC 7763, meanwhile, includes a default thermal design power (TDP) of 280 W. On an absolute scale, this translates to 4.375 W/core for EPYC 7763 and 6.75 W/core for Xion.
Arm vs. x86: Comparison to Apple’s M1 Processor
Apple’s release of a processor chip marked its transition from Intel silicon to its own in-house integrated processors. The Arm-based M1 is built on a 5nm process technology and unifies multiple chipsets into an SoC architecture, including CPU, GPU, and neural engine.
Compared to Intel’s processor, which is an x86 architecture, it comes with an 8-core CPU that the company claims consists of four high-performance cores and four high-efficiency cores. With up to 8 GPU cores, Apple states that M1 can execute nearly 25,000 threads at a time and a 16-core neural engine that runs 11 trillion operations per second.
Apple’s M1 processor architecture. Image used courtesy of Apple
Apple’s transition from Intel’s x86-based processor to its in-house Arm-based M1 processor allows the company to have higher quality assurance for hardware and more control over its IP.
Product Application Matters
Intel and AMD’s processors are based on x86 architecture with varying functionality, features, and technology. Intel’s latest offering seemingly delivers a strong appeal for data centers with AI workloads as well as high-performance computing (HPC).
However, AMD is also creating an extremely competitive environment that ultimately boosts the processing capabilities required for cloud, HPC, networking, AI, and enterprise workloads. Single thread performance and multi-thread performance are popular metrics to quantify processor performance. Cost per thread, in turn, also becomes an important parameter driving the adoption of such processors.
While these x86 processors are oriented toward server loads, Apple’s Arm processor is dedicated to Mac computers. Arm processors have considerably lower power consumption, which suits them to Mac’s M1 processor chipset devices. On the other hand, x86 processors have historically demonstrated better performance for higher power consumption applications.
This post was first published on: All About Circuits