Altia Discusses Low Cost Graphics Hardware Options in Embedded Computing Design
Friday, December 24th, 2010In the August issue of Embedded Computing Design magazine, Peter Abowd, Altia President of Worldwide Automotive, and Jim Mikola, Senior Engineer for Altia, discuss inexpensive graphics hardware options and present a case for each. In “Contrasting sprites and GPUs and the HMI modeling approach.” Abowd and Mikola discuss the best scenarios for sprite and GPU and explain the challenges that design teams must consider when implementing these hardware options into their products.
This discussion is timely, as embedded displays are becoming pervasive – especially in automobiles. Automotive companies want to deliver exciting and media-rich instrument clusters and center stacks in their new products, but they must deliver these systems at a cost that will suit their bottom line. GPUs have been lighting up automotive displays for several years; the less expensive sprite controllers are new to automotive applications, so their capabilities might not be well understood. When making hardware selections for their new products, engineers and product managers need to carefully match their product requirements to their bill of materials – so being informed about inexpensive alternatives that are new to the game is critical. Ultimately, they want to deliver the most power and pizzazz at the lowest cost.
Because Altia is not tied to any semiconductor companies or technologies, and its tools are not constrained to one operating system, the company freely supports both technologies. Altia’s tools are proven with sprites and GPUs. Therefore, the goal of this discussion – and Altia’s mission as an automotive HMI engineering leader – is to help automotive designers get the best HMI running in their products on the best possible mechanism for their applications.
What are sprites and GPUs?
New to automotive display applications, a “sprite”, or sprite-based graphical controller, is semi-conductor chip whose graphical features and capabilities have been optimized to display previously created graphical images, such as icons, often referred to in computer graphics terms as sprites. These specialized types of graphical display controllers build full screen scenes from assembling the predefined pieces, or sprites, into specific locations. Different semi-conductor manufacturers provide varying additional features such as scaling, or blending the sprites, but all essentially require the image, or sprite, to be an already created, static image. Sprite chips assemble a screen each time the frame is clocked to the display (typically 60 times a second).
A graphics processing unit (GPU) renders its display image from a set of primitive graphical operations, which can be quite computationally intensive. A GPU would render a rectangle using line draw primitives, and apply a surface to the rectangle through various other methods and calculations. These are mathematically-intensive tasks, which otherwise, would put quite a strain on the CPU. The GPU lifts this burden from the CPU, freeing up cycles that can be used for other jobs. GPU chips assemble their screens only when the screen is constructed and the resulting framebuffer is clocked to the display.