An introduction to AWS Graviton processors and instances

An Introduction to AWS Graviton Processors and Instances

AWS Graviton processors are custom-built by Amazon Web Services (AWS) to deliver the best price performance for cloud workloads[1]. These processors represent a significant shift in cloud computing, offering an alternative to traditional x86-based processors from Intel and AMD[1][2]. AWS Graviton processors are based on the Arm architecture, providing efficiency, scalability, and reduced costs[2][4]. Since the initial launch in 2018, AWS has released multiple generations of Graviton processors, each offering improvements in performance and efficiency[1][5][6].

What are AWS Graviton Processors?

AWS Graviton processors are 64-bit Arm-based CPUs designed by AWS, utilizing the expertise of its subsidiary, Annapurna Labs[2][4]. These processors are specifically optimized for cloud workloads running in Amazon Elastic Compute Cloud (EC2)[4]. The primary goal is to provide high performance, scalability, and reduced latency for various applications[2][1]. Graviton processors are not sold outside of Amazon and are designed to be tightly integrated with AWS servers and data centers[2].

Why Did AWS Create Graviton Processors?

AWS created Graviton processors to offer customers more choice in selecting EC2 instances[2]. By targeting Arm-based applications, AWS aimed to provide high availability and security while reducing virtualization costs[2][4]. The development of Graviton processors stems from AWS’s acquisition of Annapurna Labs in 2015[4]. This partnership initially focused on the AWS Nitro System, which integrated specialized hardware with a lightweight hypervisor to maximize hardware power for EC2 instances[4]. Following the success of Nitro, Annapurna Labs developed its custom CPUs to efficiently handle scale-out workloads[4].

The benefits AWS anticipated included:

* **More Choice:** Offering a wider selection of EC2 instances for customers[2].
* **Arm-Based Applications:** Targeting and supporting applications optimized for the Arm architecture[2][3].
* **High Availability and Security:** Ensuring robust performance and security for cloud workloads[2][4].
* **Reduced Virtualization Costs:** Lowering the costs associated with virtualization through optimized hardware[2].
* **Improved Performance and Lower Prices:** Delivering decent server performance at more competitive prices[2].

Benefits of AWS Graviton

AWS Graviton processors offer several key advantages:

* **Price Performance:** Graviton processors provide better price performance compared to traditional x86 processors for many cloud workloads[1][4].
* **Lower Energy Use:** Graviton processors consume less energy relative to x86-64 CPUs, contributing to more sustainable cloud operations[2][6].
* **Scalability:** These processors allow for efficient scaling of applications, ensuring consistent performance as demand grows[1].
* **Reduced Latency:** With optimized architecture, Graviton processors minimize latency, providing faster response times for applications[1].
* **Cost Efficiency:** Organizations can reduce overall computing costs due to the processors’ efficient design[1][4].
* **Effective CPU Power:** AWS Graviton processors offer competitive performance compared to traditional architectures[3].
* **General Purpose Use:** Built to improve efficiency in servers, mid-size data-storing processes, micro-services, and cluster computing[3].
* **Burstable Workload:** Providing an extensive set of burstable workload services such as scale microservices, small and medium database services, virtual desktops, and a selection of applications suitable for critical business[3].

AWS Graviton Generations

AWS has introduced four generations of Graviton processors in the last five years[5]:

* **Graviton (2018):** The first generation of AWS Graviton processors, launched in 2018, powered general-purpose A1 EC2 instances[2][4]. These instances provided up to 16 vCPUs and paved the way for the software ecosystem needed for subsequent generations[5].
* **Graviton2 (2019):** Introduced in 2019, Graviton2-based instances offer up to 40% better price performance compared to fifth-generation instances[1][5]. Graviton2 processors feature large L1 and L2 caches for every virtual central processing unit (vCPU), ensuring workloads fit in the cache without needing to access memory[1]. Each vCPU is a physical core, providing isolation and preventing resource sharing[1]. Cores are connected in a mesh with ~2TB/s of bisection bandwidth, allowing applications to move quickly between cores when sharing data[1]. Graviton2 instances include up to 64 vCPUs and use Arm Neoverse N1 cores[5].
* **Graviton3 (2021):** Graviton3 processors deliver 25% better performance and 50% faster memory access compared to Graviton2, along with 60% lower energy consumption[6]. Graviton3 offers up to 2x better floating-point performance, up to 2x faster crypto performance, and up to 3x better ML performance compared to Graviton2[5]. Graviton3 uses Arm Neoverse V1 cores[5].
* **Graviton4 (2023):** Announced in late 2023 and made generally available in July 2024, Graviton4 provides up to 30% better compute performance compared to Graviton3[6]. Graviton4 increases the core count to 96 and will be first available in the R8g instance type[5]. Graviton4 provides 30% better compute performance, 50% more cores, and 75% more memory bandwidth than Graviton3[5].

AWS Graviton Instance Types

AWS Graviton powers various Amazon EC2 instance types suited for general-purpose, compute-optimized, memory-optimized, and storage-optimized use cases[1]. These instances are available in most AWS Regions, including GovCloud and AWS China Regions[1]. There are more than 150 instance types with Graviton processors[5].

Key AWS Graviton Instance Families

* **General Purpose:**
* **T4g:** Provides a balance of compute, memory, and networking resources for a wide range of general-purpose workloads.
* **M6g and M7g:** Designed for general-purpose workloads that require a balance of compute, memory, and networking.
* **Compute Optimized:**
* **C6g and C7g:** Optimized for compute-intensive workloads, such as high-performance computing (HPC) and video encoding.
* **Memory Optimized:**
* **R6g, R7g, and R8g:** Designed for memory-intensive applications, such as databases and in-memory caches.
* **Storage Optimized:**
* **Im4g and Is4g:** Optimized for I/O-intensive workloads that require high sequential read and write access to large datasets.

When to Use AWS Graviton

AWS Graviton processors are suitable for a wide range of workloads, including:

* **General-purpose applications:** Web servers, application servers, and microservices[3].
* **Compute-intensive tasks:** High-performance computing (HPC), media encoding, and scientific modeling[3].
* **Memory-intensive workloads:** High-performance databases and real-time big data analytics[3].
* **Arm-based applications:** Applications specifically designed and optimized for the Arm architecture[2][3].
* **Mid-size Data-Storing Processes:** Efficient for handling medium-sized data storage needs[3].
* **Cluster Computing:** Suitable for distributed computing environments[3].
* **Gaming:** Built on a computer-intensive model like gaming[3].
* **CPU-based computer learning processes:** Useful for CPU-based machine learning tasks[3].

Performance Improvements with Graviton

Since the introduction of the A1 instance type, performance has increased by about 4x per vCPU, and the core count has increased 6x[5]. Graviton2 offers 35% faster performance running Redis, 30% faster running Apache Cassandra, and up to 117% higher throughput for MongoDB[2]. In addition to higher performance, Graviton offers 70% lower power consumption and 20% lower price[2].

Conclusion

AWS Graviton processors provide a compelling alternative to traditional processors, offering enhanced price performance, scalability, and energy efficiency[2][4]. By utilizing Graviton processors, organizations can optimize their cloud workloads and reduce costs[1]. As AWS continues to innovate with its Graviton processors, users can expect further advancements in performance and efficiency, making Graviton an increasingly attractive option for a wide range of cloud-based applications[5][6]. With each new generation, AWS Graviton processors are solidifying their position as a key component in modern cloud infrastructure[5].