
W9971CF
5
FUNCTIONAL DESCRIPTIONS
Graphics Engine
The 64-bit graphics engine (GE) is specially designed to improve the performance of Microsoft
Windows applications and other graphical user interface software. Performance is enhanced by
accelerating GUI functions such as BitBLTs, Bresenham line draw, and short stroke vector. The GE
performs several types of BitBLTs, including HostBLT, pattern BLT, color/font expanding BLT,
transparent BLT, color expansion, and rectangle fill, with 256 ROPS on source, pattern, and
destination. The graphics engine operates at all pixel depths in enhanced modes, including 8/16/24
bit-per-pixel modes. A fully Microsoft Windows compatible 64
×
64
×
2 or 32
×
32
×
2 hardware
cursor, is used to speed up cursor performance and accelerate all graphics operations by eliminating
the CPU overhead associated with cursor manipulation.
Video Accelerator
The Video Accelerator is a high-quality video processor for enhancing and displaying video images.
It not only accelerates software video playback but also facilitates the interface to MPEG-1 video
decoders and TV decoders.
The video accelerator employs back-end color space conversion and a scaling mechanism. Video
and game-generated graphics frames are read from an off-screen buffer and sent to the overlay and
blending control block as the secondary stream. Color space conversion and scaling are performed
along the way. This on-the-fly video acceleration and overlaying ensure game and video playback at
full speed.
Color Space Conversion
The built-in color space converter converts YUV format to 24-bit RGB format according to the CCIR
601-2 Recommendation.
Down Scaling
The down-scaling function is performed by dropping pixels or lines. To support arbitrary scaling
factors, the video accelerator uses a digital differential accumulator (DDA) to drop pixels or lines
smoothly. The scaling factor may range from 1/64 to 63/64.
Scaling up by 2D Interpolation
Rather than duplicating pixels or lines for enlargement, the video accelerator uses a 2-D bilinear
interpolation algorithm to enhance the output for high-quality video display.
The interpolation algorithm creates seven weighted, intermediate values between two original pixels
or lines. This allows a maximum enlargement of 800%. For non-integer scaling, a 10-bit DDA with
1/1024 resolution is used to interpolate pixels by averaging.
2x Scaling
The 2x scaling located in the RAMDAC is used to optionally scale up the overlaid data stream by a
factor of 2. This is used to scale 320
×
240 game-generated frames to full screen.