Jump to content United States-English
HP.com Home Products and Services Support and Drivers Solutions How to Buy
» Contact HP
More options
HP.com home
HP PEX Implementation and Programming Supplement: HP9000 Series 700 Color Workstations > Chapter 5 Performance Hints

Identify SPU and Graphics Hardware Suited For the Application

» 

Technical documentation

Complete book in PDF
» Feedback
Content starts here

 » Table of Contents

 » Glossary

 » Index

Your choice of a hardware platform will depend on the type of application you are planning to support. For example, will end users spend only a small amount of time creating a model, but spend most of their time rotating objects and viewing them from different directions? If so, the graphics hardware may be the more important consideration. On the other hand, if the application is going to solve complicated equations while rendering, the choice of CPU may be more important to your application performance.

Where to Get Information About HP Systems

Benchmarks and technical information about HP systems is published in the Product Data Sheet for that system, which is available from your sales representative. If you have access to World Wide Web, you can also find much of this information by opening Hewlett Packard's home page, at http://www.hp.com. From the home page, you can access information about HP computers and peripherals, support services, hints for troubleshooting problems on HP systems and other timely information. Another source of news about HP products is The Hewlett-Packard Journal.

System Level Benchmarks

Several benchmarks are published about HP graphics systems that should help you to determine if the system is an appropriate choice for your application. Typical system data includes SPECint and SPECfp ratings, and Linpack and Dhrystone benchmark results.

Graphics Benchmarks

The Graphics Performance Characterization committee (GPC) provides a set of Picture Level Benchmarks (PLBs), which are a standardized, vendor-independent measure of graphics performance. The benchmarks are run from the PLB interpreter program, which executes a series of graphics calls. A number of data sets, which contain graphics calls that a typical application might make, are provided by the GPC committee. For example, PLBWire93 is a good indicator of wireframe application performance. PLBSurf94 is a good indicator of 3D shaded surface performance. Xmark93 is a good indicator of how the user interface will perform.

Other published graphics data includes triangles/second, vectors/second and quadrilaterals/second. This type of performance number is not always available for a given device. Beware of these specifications, because they rarely reflect actual performance of an application. These benchmarks and PLBs are usually available on the Product Data Sheet. Other information about GPCs is available in the GPC Quarterly. You can receive copies of that publication from university libraries, by subscription, or from a sales representative. The GPC Quarterly is published by the National Computer Graphics Association (NCGA), 2722 Merrilee Drive, Suite 200, Fairfax, VA 22031.

Other Considerations

Other performance considerations for 3D graphics users include:

  • Does the graphics system include a hardware Z buffer? A hardware Z buffer is used to accelerate hidden surface removal. If your application renders 3D solids or surfaces, a hardware Z buffer will accelerate rendering and animation of complex models.

  • Is hardware double-buffering supported at the depth your application needs? Double-buffering allows smooth movement of dynamic images. To the human eye, double-buffered animation sequences appear to run faster.

  • Does the system include overlay planes? Running the graphical user interface (GUI) in the overlay planes and the graphics in the image planes can result in a substantial performance improvement for some 3D applications. This is because exposure events, caused by GUI interactions like pop-up menus, can force expensive redrawing of the graphics images when the pop-up menus disappear.

  • Is rasterization of primitives accelerated? If it isn't, rendering times are strongly impacted by window sizes.

  • How does the graphics system interact with the CPU? If the graphics computations are done completely by the graphics device, graphics performance will not scale with a faster CPU. If the mathematical computations are done in the CPU, instead of by specialized graphics hardware, graphics performance will scale with CPU performance.

  • Is texture mapping supported in hardware? Texture mapping gives an application the ability to map a 2D image onto a 3D surface for a more realistic rendering.

Software double-buffering, hidden surface removal and texture mapping are supported on all HP graphics devices. However, hardware support significantly improves performance for this functionality.

Printable version
Privacy statement Using this site means you accept its terms Feedback to webmaster
© 1996 Hewlett-Packard Development Company, L.P.