Dell is rolling out five new AMD-based servers to enhance its AI Factory, which focuses on infrastructure and services for AI tasks. The lineup features four models powered by the fifth-generation AMD EPYC Turin processors: the R6715 and R7715 are single CPU models, while the R6725 and R7725 come with dual CPUs. The fifth model, the XE7745, also runs on the new AMD CPUs and supports up to eight double-width or 16 single-width GPUs, aiming at AI inferencing.
In addition to the hardware, Dell has introduced version 10 of its integrated Dell remote access controller (iDRAC) software, which now includes 256-bit encryption for improved security. They’ve rolled out the Dell Generative AI Solutions with AMD, which combines infrastructure, software, and professional services. Dell is also deepening its collaboration with Hugging Face, an open-source platform for model training, to incorporate AMD hardware options.
This shift to AMD makes sense; their CPUs deliver higher performance at a lower cost compared to Intel, according to Anurag Agrawal, CEO of Techaisle. AMD also provides better performance per watt, which is crucial for customers focused on cost optimization and return on investment in AI.
Although these new servers broaden Dell’s AI Factory offerings, clarity on how customers can utilize them effectively is still needed. Agrawal pointed out that while the factory concept provides foundational components, it lacks a clear roadmap to turn raw materials into finished products.
The AMD-based R-series servers feature modular designs that enhance air cooling, which benefits the dual 500W AMD CPUs. Analyst Camberley Bates noted that this architecture is a big leap for energy efficiency and performance density. These servers don’t just cut CPU costs, but also reduce energy expenses relative to their performance. Data centers face increasing power and cooling demands for AI applications, and this puts server vendors like Dell in a competitive position.
Dell has also reported record performance in VMmark 4 (virtualization) and TPCx-AI (machine learning) benchmarks. The PowerEdge XE7745 is tailored for AI inferencing and model tuning, designed for data-heavy workloads. The XE or “extreme acceleration” series targets demanding applications like AI and high-performance computing. While the XE9680, released in mid-2023, uses Intel CPUs, the XE7745 shifts to AMD CPUs, coupled with GPU slots and PCIe Gen 5 connectivity. This model is suited for on-premises AI workloads, especially for regulated industries like finance and healthcare that can’t rely on cloud solutions for things like inferencing.
Dell’s AI Factory also encompasses professional services, including Kubernetes setups and AI deployment frameworks, now extended to AMD environments. Agrawal highlighted Dell’s commitment to customer choice in their go-to-market strategy. Their professional services teams offer Accelerator Workshops, allowing customers to select options based on cost and deployment timelines.
While the PowerEdge XE9680 has supported AMD accelerators since its launch, the newer offerings integrate into the broader AI Factory ecosystem, including Dell Generative AI Solutions with AMD and the partnership with Hugging Face. The Hugging Face-Dell portal will now allow the use of AMD Instinct MI300X accelerators for on-premises model deployment. Bates emphasized that this is part of Dell’s effort to simplify and streamline on-premises system deployments.