Abstract; Introduction; Visualization; Graphics Processor Technology - HP 234664-002 - ProLiant - ML330T02 Technology Brief

Abstract

Visualization and Acceleration are two key segments of the compute-intensive applications market
today. This technology brief provides an overview of Visualization and Acceleration technologies and
products for industry-standard servers and their primary applications areas. It describes the features
that are being designed into HP ProLiant servers to support Acceleration and Visualization products,
and it describes the ongoing qualification and testing efforts undertaken at HP to ensure that these
products provide robust and optimized solutions when used with HP ProLiant servers. This paper is
intended for IT professionals who are aware of Visualization and Acceleration technologies and are
potentially interested in either using or developing compute-intensive solutions that employ them.

Introduction

Visualization products are designed specifically to speed up graphics-based calculations. They are the
next step beyond the smaller graphics engines found on many graphics cards today. Visualization
products are often referred to as Graphics Processing Units (GPUs).
Accelerators are intended to speed up the processing of math calculations in general, not just
graphics calculations. Many accelerator products are re-purposed GPU's that have added a front-end
Applications Program Interface (API) and development system for programming mathematics-intensive
applications.

Visualization

Visualization is a general term for applications that process and render large amounts of data into a
graphically displayed format using modern graphics controllers.

Graphics processor technology

At the heart of all modern graphics controllers are GPU's, processor chips that are specifically
designed to perform the complex mathematics required for rendering 3D graphics. Visualization
applications typically use one of two major graphics APIs, Microsoft DirectX or OpenGL, to program
the graphics commands that are processed by the GPUs to generate screen images.
The 3D graphics rendering involved in visualization applications is a process that lends itself well to
being parallelized, and GPU's typically consist of many processor cores that operate in parallel to
perform the millions of graphics operations involved in the rendering process. Today's high-end
graphics processors for visualization applications are really the professional counterparts of the
graphics cards used in consumer applications such as PC gaming. These graphics cards feature GPUs
that may contain several hundred separate processor cores on a single chip and are capable of
executing hundreds of millions graphics operations per second.
2