‘Our customers want reduced time to value. They want to be able to deploy AI faster and get value out of it quicker. We’ve been hearing over the last two years that it takes a lot of effort to actually deploy AI solutions and get them to work,’ says Dell Technologies Senior Vice President of Product Marketing Varun Chhabra.

Dell Technologies has partnered with AMD to build a new lineup of PowerEdge Servers that is setting world records in benchmark testing while offering customers more cores and better performance.

“One of the things we’ve been focused on at Dell is really reducing that time to value,” Dell Technologies Senior Vice President of Product Marketing Varun Chhabra said in a media pre-briefing announcing the new servers. “Our customers want time to value. They want to be able to deploy AI faster and get value out of it quicker. We’ve been hearing over the last two years that it takes a lot of effort to actually deploy AI solutions and get them to work.”

The 4U PowerEdge XE7745, double-socket PowerEdge R6725 and R7725, and single-socket PowerEdge R6715 and R7715 are the newest additions to Dell’s server lineup, promising more compute with less power consumption.

Dell said one of the new AMD EPYC-backed servers achieves the same computing power as seven of its servers from five years ago while using 65 percent less power. AI data center power consumption is 10 times that of traditional data centers, leading those providers scrambling for capacity and hunting for ways to conserve electricity.

“They are seeking opportunities to consolidate so that they can simplify and streamline their data centers, their ecosystems for AI, so that they can enable themselves to operate with greater speed, more agility,” Chhabra said.

Each of the new PowerEdge servers runs the AMD EPYC processor and supports up to 50 percent more cores with up to 37 percent increased performance per core, resulting in greater performance, efficiency and improved TCO, said Arunkumar Narayanan, senior vice president of server and networking products at Dell.

“What we’ve done here is we’ve built a brand-new chassis with DC-MHS designs that will enhance our air-cooling and we can do 192 cores per CPU and 500 watts CPUs for unmatched power and efficiency,” Narayanan said on a call with media. “We’ll be industry-leading in the ability to do air-cooling in a 2U capability with 500 watts. In addition to that, we are going to have richer storage configurations. So a lot of the density-optimized storage options we are going to enable in this server.”

“With this server launch we are going to hit a bunch of world records,” Narayanan said.

The new servers are ranked first in the VMMArk4, which measures the performance and scalability of virtual environments, as well as the TCPx-AI benchmark, which measures the performance of an end-to-end machine learning or data science platform.

Dell PowerEdge XE7745

Designed for enterprise AI workloads, the XE7745 can hold up to eight double-width GPUs or 16 single-width PCIe GPUs and support power-hungry 600-watt GPUs in a 4U, air-cooled chassis. This compares with the XE9680, which Narayanan said is the company’s most successful product, reaching $10 billion in sales alone.

He said the XE7745 is purpose-built for AI inferencing, model fine-tuning and high- performance computing. The internal GPU slots are paired with eight additional Gen 5.0 PCIe slots for network connectivity, creating dense, flexible configurations with 2X more DW PCIe GPU capacity.

“We expect this to be the mainstream platform for enterprise inferencing solutions as we get into the next generation of enterprise AI adoption,” Narayanan said.

The Dell PowerEdge XE7745 server will be available globally starting January 2025.

PowerEdge R6725 And R7725

The two-socket servers are built for scalability with high-performing AMD 5th generation EPYC processors. The new DC-MHS chassis design gives it enhanced air-cooling and dual 500W CPUs.

Narayanan said Dell is among the first to put that much compute into a 2U air-cooled chassis.

He said the R6725 and R7725 maintain rigorous data analytics and AI workloads, with configurations optimized for scalability. The R7725 offers up to 66 percent increased performance and up to 33 percent increased efficiency.

The servers will be available globally starting November 2024

PowerEdge R6715 And R7715

Dell rounded out its server announcements with the 1U PowerEdge R6715 and R7715, which come equipped with AMD 5th generation EPYC processors. The design has up to 37 percent increased drive capacity, resulting in greater storage density.

Dell said it is available in various configuration options. The single-socket servers support double the memory with support for 24 DIMMs (2DPC) and meet diverse workload requirements and maximize performance in compact 1U and 2U chassis.

The servers will be available globally starting November 2024

Dell AI Factory Enhancements To Speed GenAI Deployments

Dell AI Factory, which debuted earlier this year, offers an end-to-end lineup of products and services needed to deploy AI within a business and now comes in an AMD flavor. Dell partners with Hugging Face for the software layer.

For on-premises GenAI, the Dell PowerEdge XE9680 server with AMD Instinct MI300X accelerators can deliver inferencing, retrieval augmented generation (RAG) and customization. Dell said the deep integration of AMD into the product improves security and performance, reduces time to value by up to 86 percent, and helps organizations optimize their AI investments.

Inside the factory, the Dell Enterprise Hub on Hugging Face provides custom containers and scripts for easy, safe deployment of AI models such as Llama and Mixtral. These containerized models are uniquely optimized to boost inferencing performance based on the model and server and leverage the Hugging Face Text Generation Inference back end.

Professional Services For GenAI

Dell professional services for GenAI are growing and now support AMD environments. Dell said its implementation services for GenAI platform with AMD gives customers a tailored operational platform, including Kubernetes configuration, deployment of advanced AI frameworks and best practices.