White Papers

Extreme GPU Computing
Page 3
1. Introduction
GPU computing is now established and widespread in the HPC community. There is an ever-increasing
demand for compute power. This demand has pushed server designs towards higher hardware
accelerator density. However, most such designs have a standard system configuration, which may not
be optimal for maximum performance across all application classes. The latest high-density design
from Dell, the PowerEdge C4130, offers up to four GPU boards in a 1U form factor. In addition, the
uniqueness of PowerEdge C4130 is that it offers a configurable system design, potentially making it a
better fit, for the wider variety of extreme HPC applications.
In this paper, we describe the new Dell server, C4130; specifically designed for GPU Computing, and
demonstrate how a wide class of applications has seen multifold increase in performance on the C4130.
2. The PowerEdge C4130
The PowerEdge C4130 is the latest addition to Dell’s portfolio to build HPC solutions. The C4130
incorporates several innovations in its system design to enable extreme GPU computing.
Figure 1: The PowerEdge C4130 server
Unique system configurability
The C4130 offers five carefully selected system configurations. The motivation behind offering this
configuration is to offer a system design that will best fit the application requirement. This offers users
a unique opportunity to match the system design with their application characteristics. The
configurations differ in the manner in which accelerators connect to the CPUs, resulting in different
accelerator to CPU ratios. They vary from balanced to clustered configurations.
The goal of this white paper is to introduce available server configurations. The C4130 offers five
configurations shown in Figure 2 and 3, A through E. Since GPUs provide the bulk of compute
horsepower, the configurations can be divided into groups based on expected performance. The first
group of three configurations, A, B and C, has four K80s each. The second group of two configurations,
D and E, has two K80s each. The first two quad GPU configurations have an internal PCIe switch
module. The details of the various configurations are shown in the Table 1.