In a significant step towards innovation, Google has announced its latest custom silicon with the introduction of Google Axion Processors, the company's first Arm®-based CPUs designed explicitly for data centre deployment. This development shows Google's goal of pushing computing boundaries to serve grand challenges like information retrieval, global video distribution, and generative AI.
Google's History of Custom Silicon:
Google's journey into custom silicon dates back to 2015 with the inception of Tensor Processing Units (TPU), followed by advancements like Video Coding Unit (VCU) in 2018 and Tensor chips for mobile devices in 2021.
These investments have changed customers' capabilities and addressed the need for enhanced computing power, particularly in analytics, information retrieval, and machine learning (ML) training and serving.
Axion Processors: Unrivaled Performance and Efficiency:
Axion processors use Google's expertise in silicon design and Arm's highest-performing CPU cores, promising up to 30% better performance than existing Arm-based instances and up to 50% better performance and 60% better energy efficiency than current-generation x86-based instances.
Axion is built on Arm Neoverse™ V2 CPU. Axion processors offer significant performance enhancements across various general-purpose workloads, including web servers, containerized microservices, data analytics engines, media processing, and CPU-based AI training and inferencing.
Underpinned by Titanium for Enhanced Efficiency:
Axion processors are underpinned by Titanium, a system comprising purpose-built custom silicon microcontrollers and tiered scale-out offloads, optimizing platform operations like networking and security.
Titanium offloads storage I/O processing to Hyperdisk, a new block storage service, thus enhancing performance and capacity for customer workloads while contributing to energy efficiency.
Application Compatibility and Interoperability:
Axion processors are built on the standard Armv9 architecture and instruction set, ensuring out-of-the-box application compatibility and interoperability.
Google's contributions to the Arm ecosystem, including Android, Kubernetes, Tensorflow, and the Go language, and collaboration with Arm and industry partners, enable seamless deployment of Arm workloads on Google Cloud.
Key Takeaways:
- Google's Axion Processors mark a significant advancement in data centre computing, offering unparalleled performance, efficiency, and compatibility.
- The integration of Arm-based CPUs underscores Google's commitment to driving innovation and addressing evolving customer needs in the cloud computing landscape.
- With Axion processors, Google aims to enhance customer experiences and contribute to sustainability goals by optimizing energy efficiency in data centre operations.
Recommended Newsletters 🐝 🐝 🐝 🐝
Source:
