Google Announces EPYC-Based Tau Virtual Machines

Google Launches Tau Virtual Machines Powered by AMD EPYC

This morning, Google unveiled Tau, a novel series of virtual machines constructed utilizing AMD’s third-generation EPYC processor. The company asserts that this new x86-compatible system delivers a 42% improvement in price-performance compared to conventional VMs.

Google initially began incorporating AMD EPYC processors into its Cloud infrastructure in 2017. Amazon Cloud followed suit, introducing its own EPYC-based offerings in 2018.

Enhanced Performance and Configuration Options

Google positions the Tau family as surpassing current cloud VMs in capabilities. These systems are available in diverse configurations, scaling up to 60 virtual CPUs (vCPUs) per VM.

Each vCPU is supported by 4GB of memory. Networking bandwidth reaches up to 32 Gbps, and the VMs are compatible with a range of network-attached storage solutions.

Executive Statement and Cloud Strategy

“Organizations across all sectors are facing increasingly complex and data-intensive workloads,” stated Thomas Kurian, CEO of Google Cloud, in a press release. “They are actively seeking methods to accelerate performance and minimize expenses.”

Kurian continued, “Our partnership with AMD has enabled us to expand our portfolio and provide customers with optimal price-performance for demanding, mission-critical applications – all within the industry’s most sustainable cloud environment.”

Early Adopters and Initial Results

Google has already secured several prominent clients for early access trials, including Twitter, Snap, and DoIT.

Nick Tornow, Platform Lead at Twitter, commented in a blog post, “Achieving high performance at a competitive price is paramount as we strive to facilitate global conversations.”

He added, “Preliminary tests indicate the potential for double-digit performance gains. We are collaborating with Google Cloud to thoroughly assess the price and performance benefits for specific compute workloads using the new Tau VM family.”

Availability and Pricing Details

The Tau VMs are scheduled for release on Google Cloud in the third quarter of this year. Pre-registration for access is currently open to clients.

Pricing will vary based on the chosen configuration. As an example, a 32vCPU VM equipped with 128GB of RAM is projected to cost approximately $1.35 per hour.