IBM, a leader in global hybrid cloud and AI solutions, unveiled the specifications of its new Telum II processors and Spyre AI accelerators, which are set to drive AI workloads on the latest IBM Z mainframe systems.
The announcement was made at the Hot Chips 2024 event happening this week at Stanford University.
As GenAI initiatives transition from proof of concept to production, there is an increasing demand for power-efficient, secure, and scalable solutions. According to research from Morgan Stanley, GenAI power demands will rise 70% annually for the next few years. The research indicates that by 2027, GenAI can consume as much energy as Spain needed to power itself in 2022.
The new Telum II process and Spyre AI accelerators are engineered to address these escalating demands effectively.
The Telum II processor builds on its predecessor, the first-generation Telum chip, with several key improvements. The new processor features a completely new data processing unit (DPU) designed to accelerate complex IO protocols for networking and storage on the mainframe and to improve key component performance.
IBM claims the new DPU offers increased frequency, memory capacity, and an integrated AI accelerator core. This allows it to handle larger and more complex datasets efficiently.
The Spyre accelerator complements the Telum II processor by offering additional AI compute capabilities and supports what IBM calls “ensemble methods of AI modeling” - which combines multiple models to potentially boost prediction accuracy.
Ensemble methods enhance the robustness of AI predictions, making them more reliable and less sensitive to errors or variations compared to individual models. The IBM Spyre accelerator chip will be delivered as an add-on option. Each Spyre chip contains 32 compute cores for AI applications, which reduces latency and enhances throughput across various AI tasks.
"Our robust, multi-generation roadmap positions us to remain ahead of the curve on technology trends, including escalating demands of AI," said Tina Tarquinio, VP, Product Management, IBM Z, and LinuxONE.
"The Telum II processor and Spyre accelerator are designed to deliver high-performance, secured, and more power-efficient enterprise computing solutions. After years in development, these innovations will be introduced in our next-generation IBM Z platform so clients can leverage LLMs and generative AI at scale."
The new Telum II processor is a major upgrade to the original Telum processor that debuted in 2021. With 8 high-performance cores running at 5.5 gigahertz and with 36 megabytes of memory per core, Telum II processors offer an increase of 40% in on-chip cache capacity for a total of 360MB. The integrated AI accelerator enables low-latency, high-throughput AI inferencing during transactions.
The powerful specifications of the two new technologies translate to better efficiency and security for AI-powered applications. While the Telum II processor is designed to efficiently manage large-scale AI workloads and data-intensive business needs, the Spyre accelerator is geared toward handling complex AI models and generative AI use cases.
The integration of Telum II and Spyre accelerators eliminates the need to transfer data to external GPU-equipped servers, thereby enhancing the mainframe's reliability and security.
Both technologies will be manufactured by IBM's long-standing fabrication partner, Samsung Foundry, using a 5 nm process. IBM expects the Telum II processor to be available to LinuxOnE and IBM Z clients in 2025. The Spyre accelerator is also expected to be available in 2025.
Related Items
Groq’s $640 Million Funding Poised to Disrupt AI Chip Market
Is the GenAI Bubble Finally Popping?
AWS and Fujitsu Expand Partnership to Modernize Legacy Cloud Applications
