Benchmarking solution for rapid deployment of AI infrastructure
Accelerating innovation in AI/ML network validation and optimization

Keysight Technologies (hereinafter Keysight) has launched an AI data center test platform that significantly enhances benchmarking capabilities for new AI infrastructure.

Keysight announced on the 16th that it has launched the Keysight AI Data Center Test Platform, designed to accelerate innovation in AI (artificial intelligence) and ML (machine learning) network validation and optimization, while expanding its presence into the AI and ML infrastructure ecosystem.

The adoption rate of AI deployment and usage is rapidly increasing across all industry sectors, and training and deploying new AI models quickly and efficiently has become a top priority for enterprises.

AI/ML workloads process massive amounts of data, and to reduce training time requires superior networking bandwidth and computing performance.

On the other hand, the cost of conducting design and validation for large-scale "what-if" scenario assessments is prohibitively expensive, even for the largest AI operators.

Keysight's AI Data Center Test Platform overcomes these challenges and provides highly configurable AI workload emulation capabilities, pre-packaged benchmarking applications, and dataset analysis tools to accelerate the design and testing of AI/ML infrastructure, enabling significant performance improvements in AI/ML cluster network fabric.

Ram Periakaruppan, Vice President and General Manager of Keysight's Network Test and Security Solutions division, stated, "Keysight continues to lead innovation and has partnered closely with hyperscalers to co-design an AI test platform that realistically benchmarks and emulates large-scale AI workloads."

To accelerate AI/ML network design, Keysight's data center solution provides deep insights into collective communication performance through large-scale AI workload emulation leveraging measurable accuracy, and offers AI network fabric validation capabilities through pre-packaged benchmark applications developed in partnership with the largest AI operators and AI infrastructure companies.

Additionally, it supports experiment reproducibility through sharing among users and customers by executing defined AI/ML behavioral models, and offers multiple test engines to enable users to choose between AI workload emulation using Keysight hardware load appliances and software endpoints of actual AI accelerators to compare benchmarking results.

Using Keysight's platform enables practitioners to conduct large-scale validation and experimentation in a practical and cost-efficient manner related to fabric design.

This solution complements GPU-based AI/ML workload testing, providing AI operators with an integrated AI test platform with greater scalability and robustness.

Alan Weckel, founder and principal analyst at 650 Group, stated, "800GE is a key driver of AI/ML growth and the technology can have the fastest ramp relative to all data center port speeds. The growth rate of AI/ML bandwidth is forecast to exceed 100% annually through 2030, and 800GE will play an important role in supporting this growth."

He added, "Keysight's solution, which leads the AI/ML infrastructure testing field, is expected to reach production stage for multiple deployment targets within the next 18 months."

Martin Hull, Vice President of Cloud and Platforms at Arista Networks, stated, "It is already proven that high-speed Ethernet networks provide the economic scale and performance necessary to reduce AI/ML workload processing time while adhering to open standards. Keysight's AI test platform is a valuable tool for validating Arista's technical leadership in the AI network design field."