September 24, 2022

COSFONE

Networking, PBX, IT, DIY Solution

Why do more and more companies choose ARM-Architecture servers?

7 min read

Why do more and more companies choose ARM-Architecture servers?



Why do more and more companies choose ARM-Architecture servers?

There are 180 billion ARM-Architecture processors in mobile phones, IoT sensors and other devices.

Nowadays, the technology has evolved to support hyperscale data centers and cloud computing.

ARM-Architecture processors are attractive to cloud customers because of their linear scalability, predictable performance, and high core density and economics.

With ARM-Architecture cloud computing, customers can run existing workloads at a lower cost and build new applications with superior economics and performance.

Why do more and more companies choose ARM-Architecture servers?


CPU instruction sets:

The most important thing for a CPU is the core. The degree of digestion, absorption and innovation of the instruction set determines the credibility of innovation.

CPU is mainly composed of controller, arithmetic unit, memory and connection bus. Among them, the controller and the arithmetic unit form the core of the CPU.

The core extracts data from the memory, decodes the data according to the instruction set in the controller, and obtains the result through the operation of the micro-architecture (circuit) in the arithmetic unit. The execution result is written to memory.

Therefore, the foundation of the kernel is the instruction set (instruction set architecture) and the microarchitecture. The instruction set is the collection of all instructions, the instruction set can be extended (such as from 32-bit to 64-bit), it specifies all the operations that the CPU can perform.

At present, the mainstream instruction sets in the market are the complex instruction set (CISC) represented by x86 and the simple instruction set (RISC) represented by ARM.

The micro-architecture is the circuit design that completes the operations of these instructions. The same instruction set can have different micro-architectures.

For example, Intel and AMD are both based on the X86 instruction set but with different micro-architectures.

Therefore, it can be seen that the more an enterprise digests, absorbs and innovates the instruction set architecture, the more likely it is to achieve innovation.

ARM is a reduced instruction set

ARM is a reduced instruction set family that is expected to iterate an architecture version every 8 years.

The ARM architecture is a reduced instruction set (RISC) processor architecture family, which is widely used in embedded system design, featuring low cost, high performance and low power consumption, covering consumer electronics, portable devices, and computer accessories. , military facilities, etc.

After years of development, ARM has evolved from v1 to the latest v8. From ARMv3 to ARMv7, it supports 32-bit space and 32-bit arithmetic operations.

The instructions of most architectures are fixed-length 32-bit, while the ARMv8-A architecture released in 2011 Added support for 64-bit space and 64-bit arithmetic, and updated the 32-bit fixed-length instruction set.

At present, ARM architecture processors account for 75% of all 32-bit embedded RISC processors, which is one of the largest 32-bit architectures in the world.

According to ARM’s estimation, it will take about 8 years for the new generation of architecture to be popularized from research and development to commercial use.

10 years later, ARMv9 was finally released on March 31, 2021 after ARMv8 was released in October, 2011.

Why choose ARM-stucture processors?

ARM entered the server market with partners such as Ampere Computing by introducing 64-bit computing in the ARMv8 architecture and Neoverse Neoverse family.

The rise of hardware virtualization and cloud computing has accelerated this shift. In the processor industry, multi-CPU architectures will once again be the norm.

Customers will once again have the diversity and choice to build next-generation applications, with ARM, X86 and GPUs leading the way.

The ARM Neoverse N1 CPU architecture is designed for the data center and cloud infraArchitecture spaces.

It provides a high core count server-class SoC subsystem with the performance, functionality and scalability necessary to accelerate the transition to scalable cloud and edge infraArchitecture.

It fundamentally changes the performance equation for hyperscale data centers and cloud computing, enabling more work to be done with less energy than traditional technologies.

Ampere Computing’s Altra processors are based on the Neoverse N1 CPU architecture. Ampere uses the N1 platform and integrates its own innovative features to create a SoC built for applications across hyperscale cloud data centers.

N1 ARM processors have higher thread counts, higher CPU frequencies, and greater core density, which help achieve:

Breakthrough Computing Performance

Neoverse N1 ARM CPUs are specifically designed and optimized for high performance and hyperscale data centers.

These CPUs deliver sustained full-frequency performance with a single-threaded core architecture to run loads with consistent and predictable performance while achieving ideal scalability.

The cache Architecture of these CPUs is suitable for large, branch-intensive infraArchitecture workloads.

Customers can now predictably scale their loads at a lower cost. ARM processors, such as Ampere’s Altra processors, feature a single-thread-per-core architecture, allowing you to run workloads with consistent and predictable performance, while allowing ideal performance scaling.

These cores are completely isolated from the noisy neighbors of other loads running on the same processor.

‘Scale up’ workloads that require very high core counts will benefit from this, while ‘scale out’ workloads will also benefit from multiple instances of smaller VM configurations.

Predictable performance also means more predictability for month-end billing.

Significant cost savings for web servers, API gateways, media encoding, AI inference, and other processor-bound loads.

The Fugaku, the world’s fastest supercomputer jointly developed by Riken and Fujitsu in Kobe, Japan, uses Fujitsu’s ARM-Architecture system-on-chip.

ARM processors top the list of processor architectures used to build supercomputers.

Why do more and more companies choose ARM-Architecture servers?

Higher security

The Ampere Altra processor’s single-thread-per-core processor design eliminates potential thread-safety issues and provides isolation for customer workloads.

This helps reduce the risk due to side channel attacks.

By having only one thread run per core, the execution engine, registers, and L1/L2 cache are not shared between threads, minimizing the attack surface.

ARM TrustZone is a hardware security extension technology that creates an efficient system-wide security solution through hardware-enforced isolation built into the CPU.

It provides a perfect starting point for establishing a root of trust for devices based on Platform Security Architecture (PSA) guidelines.

ARM TrustZone provides a secure execution environment by splitting computer resources between two execution worlds, Normal World and Secure World.

Specifically designed for a wide range of loads

ARM server-side processors, such as the Ampere Altra, are designed to meet the most demanding workloads in a variety of data center deployments.

  • Common workloads – web applications, application servers
  • Cloud-native applications – Docker container-based applications
  • Compute-intensive applications – media encoding, AI inference, HPC applications
  • ARM native mobile app development – Android VM for developing and testing mobile apps
  • In-memory caches and databases — Redis, Memcached databases, open source databases

Why do more and more companies choose ARM-Architecture servers?


Oracle’s efforts in ARM server-side development

ARM, Ampere, and leading ISVs like GitLab and Jenkins are partnering with Oracle to make ARM server-side development easy and cost-effective.

Oracle Cloud InfraArchitecture (OCI) works with a wide range of partners to help customers transform, build, and run ARM-Architecture applications in the cloud cost-effectively.

Ampere Altra processors are based on the ARM aarch64 architecture and are now available on OCI for bare metal servers and virtual machines (VMs).

Oracle Cloud InfraArchitecture (OCI) works with a wide range of partners to help customers transform, build, and run ARM-Architecture applications in the cloud cost-effectively.

Ampere Altra processors are based on the ARM aarch64 architecture and are now available on OCI for bare metal servers and virtual machines (VMs).

Leading ISV developer tool vendors such as GitLab, Jenkins, Rancher, Genymobile, open source organizations such as the Continuous Delivery Foundation (CDF), and technology partners such as ARM and Ampere are working with Oracle to provide tools and software to build and run ARM-Architecture server-side solution.

Oracle has released Oracle Linux, Java, and MySQL distributions that support aarch64, allowing you to rapidly develop on the ARM computing platform.


Ampere Altra

The Ampere Altra processor is the industry’s first 80-core server processor, driving next-generation applications with exceptional performance, scalability and power efficiency.

  • Offers up to 160 cores in a two-socket configuration
  • Neoverse N1 Cores ARM v8 Architecture
  • No noisy neighbor issues – 1 thread/core
  • 3.0 GHZ – runs all cores at maximum frequency
  • 64 KB L1 I-cache and 64 KB L1 D-cache on each core
  • Up to 1TB RAM
  • Coherent Mesh Network (CMN)
  • AI inference acceleration
  • 7nm process technology

Ampere ARM processors are suitable for general-purpose workloads as well as rapidly evolving applications such as data analytics, mobile applications, machine learning inference, in-memory databases, web hosting and cloud-native applications.



Copyright © All rights reserved. | Newsphere by AF themes.