Speaking at the Supercomputing conference in Austin, Texas, in 2008, Dell CEO Michael Dell told the audience that his company was committed to growing its presence in the HPC space.
Dell on Sept. 9 made its latest step into the space with a host of new hardware offerings, including the PowerEdge C6105, a rack-mount server aimed at such compute-intensive environments as HPC (high-performance computing) and cloud computing.
The server is part of Dell’s C-series systems that Dell introduced in March and that are designed specifically with HPC and cloud environments in mind.
The system, powered by Advanced Micro Devices’ Opteron 4000 series processors, illustrates Dell’s approach to the highly-virtualized hyper-scale computing space, according to Donnie Bell, senior manager of HPC solutions at Dell. The IT vendor is looking to work more collaboratively with customers like Lawrence Livermore Laboratory to determine what they need in their environments, rather than create a box that is rolled out onto the market.
It’s also a high-performance computer that is affordable and manageable so that any institution can acquire it and give access to researchers and businesses-what Dell calls the “Missing Middle”-that in the past might not have had the chance to use such powerful systems.
“It was all about the elite [in the past],” Bell said in an interview with eWEEK. “So we needed to find a way to bring it to the masses.”
Not only are research institutions looking to take advantage of such servers, but enterprises now also are interested in systems that enable them to take advantage of the growing cloud-computing trend. Like HPC, cloud computing requires systems that can handle high levels of data.
The 2U (3.5-inch) C6105-which is similar to Dell’s PowerEdge C6100, which runs on Intel processors-includes four two-socket motherboards, and each socket can handle chips with up to six cores.
The system came out of Dell’s collaboration with Lawrence Livermore, which was looking for a high-density, energy-efficient system, Bell said.
The labs researchers wanted “to get a lot of compute [power] in a small space,” he said.
To help with the space and power demands, Dell made some changes to the C6105 that are not seen in its traditional rack servers, including eliminating the embedded LifeCycle Controller remote system management feature, and giving the four motherboards shared power and cooling capabilities. In Dell’s regular racks, each motherboard has its own power and cooling components.
Dell also is taking advantage of the trend toward greater use of graphics cards in the HPC space. A growing number of businesses and research institutions are beginning to use GPUs for more general-purpose computing jobs to take advantage of the parallel processing capabilities.
Dell’s new PowerEdge C410x external PCIe (PCI Express) expansion chassis can connect the C6105 servers to as many as 16 PCIe devices, including Nvidia Tesla M2050 GPUs.
Other HPC-targeted offerings include the PowerEdge M610x blade and the Precision T7500 workstation, which also is powered by Nvidia Tesla 2050 GPUs. Dell also is partnering with Platform Computing to create the Cluster Manager 2.0.1, Dell Edition, which enables users to more easily and quickly deploy and manage heterogeneous Linux clusters.
Terascala also is partnering with Dell to create standardized storage platforms for customers using the Lustre file systems.
Dell officials said it is important that the company start becoming more vocal about its HPC offerings, said Tim Carroll, senior manager for business development for Dell’s HPC unit. It’s an important business for Dell, and could be increasingly lucrative. According to Carroll, Dell estimates that between 15 to 20 percent of CPUs sold go into the technical computing environment.