MX86251 Datasheet, PDF (1/32 Pages) Macronix International – 3D graphic chipset

INDEX
MX86251
1.Introduction
A graphics subsystem using the MX86251 and 3Dfx In-
teractive Voodoo Rush^TM combines industry leading 3D
performance with the proven performance and compat-
ibility of an industry standard 2D Windows accelerator.
This union creates an extremely cost effective and un-
compromising multimedia solution. The MX86251 pro-
vides a PCI system interface and high performance VGA,
2D, and Video features in a low cost 2D chip while the
Voodoo Rush^TM delivers 3D graphics processing power.
The MX86251 connects to Voodoo Rush^TM through the
high performance 64-bit VR interface. The VR interface
supports the render/refresh operation of the 2D/3D en-
gines and the system control of the 3D devices. The
MX86251 supports not only the basic Double Buffer
Scheme, but also Triple and Quad buffer swap for super
smooth animation and 3D stereo glasses applications.
The STATUS signal provides 3D status to the PCI host
through the MX86251.
The MX86251 is based on proven MX86250 2D/Video
window accelerator technology, while adding many en-
hancements to the base functionality. All processing en-
gines on the chip are running on a faster clock, up to 75
MHz which means 600 MB/sec peak memory bandwidth.
Higher bandwidth means higher 2D performance and
higher frame rate Video. The Linear Frame Buffer write
cycle from the PCI bus is now executed in one clock cycle.
That is, the CPU now has much greater write bandwidth
into frame buffer memory. The on chip RAMDAC is en-
hanced to 160 MHz to enable many high resolution video
modes. Contrast and Brightness adjustments are added
to the Video Processor so that dark MPEG-1 videos will
look much better on the screen. ALL these features of
the MX86251 combines to make a powerful graphics
experience for multimedia.
1.1 Features
Interface to 3Dfx VOODOO RUSH^TM 3D graphic chipset
• Provide industry leading price / performance in 2D and
Video
• Enable the add-in of advanced 3Dfx Interactive
Voodoo Rush^TM 3D texture mapping and pixel
rendering engines
• 4MB frame buffer enables higher resolution 3D
graphics
• Double, Triple and Quad buffer swap
• Optimized LFB PCI write cycles with packing FIFO for
most efficient usage of frame buffer bandwidth
Very high bandwidth, up to 600 MB/sec
• Achieves single clock cycle EDO DRAM access in
Graphics Co-processor, Video Processor and Display
Processor
• Delivers 600 MB/sec bandwidth with -35 EDO DRAM
running at 75 MHz
• Provides 400 MB/sec memory bandwidth using lower
cost -50 EDO DRAM chips
High Performance 64-bit Graphics Co-processor
• High performance graphics engine with 64 bit wide data
path and memory data bus
• Uniformly accelerated graphics operations in all pixel
formats: 256 color, High color and True color
• Optimized graphics engine for BitBLTs, rectangle fill,
pattern fill, line draw, color expansion, text transfer,
and clipping
• Advanced 3 operand BitBLT ALU executes all 256
Raster Operations (ROPs)
• On chip 8x8x32 pattern memory achieves highest
throughput in the most common BitBLT in Windows
-- the Pattern BLT
• Deep on-chip Source and Destination FIFOs for
sustained burst cycles in BitBLTs
• Double buffered Co-processor registers allow
concurrent processing with CPU
• Built in hardware cursor
• Arbitrary X-Stride for efficient offscreen memory
allocation
Motion Video Codec Acceleration
• Contrast brightness adjustment
• YUV/YCrCb conversion of industry standardYUV 4.2.2
formats
• Video window zoom in both X and Y direction with
arbitrary ratio
• Interpolation with Bi-linear filters in both Horizontal and
Vertical dimensions
• Color Key supports video overlay
• Video window is double buffered
• Video is always played in True Color
Media Port interface to MPEG decoder chips or Video
Capture front-end
• Glueless interface to VMI (Video Module Interface)
P/N:PM0476
REV. 1.2 , FEB 11, 1998
1