Ultra-high density supercomputer with scalable storage. nVidia's CUDA programming API
enables developers to re-write parallel processing routines to take advantage of
high performance GPU cores. Supporting languages such as C, C++, Fortran and scientific tools such as MATLAB and
LabView, each CUDA-enabled GPU performs as a dedicated 240-thread vector processing engine.
| |
GPU Computer Cluster |
| Server Nodes |
42 Dual-CPU Nodes |
| Processing |
(2) Quad-Core Intel® Xeon® Processors per Node
Total 84 Processors - 336 CPU cores |
| Memory |
96 GB per Node - Total 4 TB RAM |
| Tesla GPU(s) |
2 GPUs (2 Teraflops) per Node
240 cores per GPU
20,160 GPU cores per 42U Cabinet |
| Connectivity |
InfiniBand, Gigabit Ethernet |
| SAN Capacity |
18TB Usable SAN Storage on High-Capacity Enterprise Disks
Dual-Head SAN & Replicated Volumes + Hot-Spares
iSCSI, FC, IB or NAS Connectivity
|
MDI's Astrix and Teras portable systems are mobile platforms for the development and execution of
GPU-accelerated applications using nVidia GeForce, Quadro and Tesla cards. MDI also offers a line of
scalable high performance computing servers based on Tesla Architecture.
In the last two decades, computer graphics have evolved from purposed 2D display devices,
to accelerated 3d rendering engines and, at last, to scientific and engineering processors.
GPUs have increased in performance many times faster than general purpose computer processors
because of the all-consuming need for more detailed, realistic, real-time 3d renderings.
And in recent years, graphics technology has finally caught up and exceeded that of traditional
high performance computing architectures used by industries such as oil & gas exploration,
scientific modeling and physics.
Leading the development of GPU computing technology is nVidia Corporation, makers of GeForce
consumer-class video cards, Quadro professional video chips and scientific & industrial-class
Tesla GPUs. Their leadership in the development of architectures based on technology previous
only used in 3d rendering has founded three generations of powerful computing engines
for all-purpose code.
nVidia developed the CUDA programming API and built a surrounding development community, which
enables traditional programming languages, such as C, C++ and Fortran to leverage the potent
computing capabilities of high-core-count vector processors on nVidia 3D rendering cards.
A CPU and a GPU act together to spin off parallel portions of an application written to take
advantage of CUDA cores through a system-level driver. Modifications to the code structure are
required to make use of those cores.
The application developer has to modify their application to take the compute-intensive kernels
and map them to the GPU. The rest of the application remains on the CPU. Mapping a function to
the GPU involves rewriting the function to expose the parallelism in the function and adding "C"
keywords to move data to and from the GPU.
The evolution of GPU computing came from the origins of 3d rendering chips.
Graphics chips started as fixed function graphics pipelines. Over the years, these graphics chips
became increasingly programmable, which led NVIDIA to introduce the first GPU or Graphics Processing
Unit. In the 1999-2000 timeframe, computer scientists in particular, along with researchers in fields
such as medical imaging and electromagnetics started using GPUs for running general purpose
computational applications. They found the excellent floating point performance in GPUs led to a
huge performance boost for a range of scientific applications. This was the advent of the movement
called GPGPU or General Purpose computing on GPUs.
The problem was that GPGPU required using graphics programming languages like OpenGL and Cg to program
the GPU. Developers had to make their scientific applications look like graphics applications and map
them into problems that drew triangles and polygons. This limited the accessibility of tremendous
performance of GPUs for science.
NVIDIA realized the potential to bring this performance to the larger scientific community and decided
to invest in modifying the GPU to make it fully programmable for scientific applications and added
support for high-level languages like C and C++. This led to the CUDA architecture for the GPU.
nVidia: "What is GPU Computing?"
Anyone can take advantage of nVidia's CUDA API by
downloading the appropriate
driver, toolkit and SDK to author their own applications. For more information on the underlying
technology of CUDA processors, see
nVidia's "Next Generation
CUDA Compute Architecture" white paper.
