MONTREAL — May 1, 1998 — Matrox® Graphics Inc. today announced the details of its new flagship MGA-G200 graphics chip and ushered in a new era of integration for 2D, 3D and video acceleration. The MGA-G200 is the first AGP graphics chip on the market to blend superior performance in every application area without compromise, creating the best all around graphics and multimedia accelerator available.
Optimized to take full advantage of today's Intel Pentium® II and next generation processors, the MGA-G200 uses a unique 128-bit DualBus architecture, up to 16MB of high-bandwidth synchronous memory and an integrated 230/250MHz RAMDAC to deliver the highest level of performance. The MGA-G200 is a full AGP 2x solution with a fast, full-featured 3D rendering and setup engine to accelerate demanding CAD and design applications, as well as 3D games. The 128-bit DualBus MGA-G200 is also pin compatible with Matrox's new value leading 64-bit MGA-G100 chip. Maintaining a pin compatibility design for its new MGA chips enables Matrox to deliver fast and cost effective solutions for integration into various system configurations.
"Just as previous generation MGA chips revolutionized the market, so too will the MGA-G200," says Dan Wood, Senior Product Manager, Matrox Graphics. "For the first time, users don't have to sacrifice performance in any application area - we've built it all into the MGA-G200."
The MGA-G200 is the first high performance graphics accelerator from Matrox to utilize a new 128-bit DualBus architecture, powerful single-ported memory and an integrated 230/250MHz RAMDAC. While dual-ported memories previously offered superior performance at high resolutions and color depths, the MGA-G200's combination of high bandwidth synchronous memory and its 128-bit DualBus architecture along with an intelligent memory controller, allow it to deliver the highest caliber application acceleration at even the most demanding resolutions and color depths. Results from Ziff-Davis' Business Graphics WinMark 98 show that the MGA-G200 achieves 160 million WinMarks at 1024 x 768 at 24-bit under Windows 95 . At this performance level, the MGA-G200 is faster in 24-bit than all its competitors running the same test in 16-bit.
The performance of the 128-bit DualBus architecture is superior to what is attainable with traditional 128-bit bus designs and far superior to any previous 64-bit design. The unique 128-bit DualBus employs two independent 64-bit buses that operate in parallel inside the graphics engine to effectively double the raw performance of most 2D operations.
The 64-bit granularity of the 128-bit DualBus means that for frequent copying of small bitmaps and fonts, performance is much higher than with a traditional 128-bit architecture. Additionally, the 128-bit DualBus architecture uses Dual Command Pipelining which permits read and write phases of two consecutive commands to be overlapped and executed simultaneously. The MGA-G200's integration of this new 128-bit DualBus architecture into a proven, high performance 2D core clearly positions the chip as the world leader in graphics acceleration.
3D Acceleration: Performance and Image Quality
The MGA-G200 integrates an extremely powerful 3D engine which delivers exceptionally high levels of performance when accelerating real world 3D applications ranging from entry-level to mid-range CAD and animation packages to next generation 3D games. The MGA-G200 accelerates both the Direct3D and OpenGL APIs and delivers hardware support for all 19 tests in Ziff-Davis' 3D WinMark 98 in a single pass.
In addition to a high performance 3D rendering engine, the MGA-G200 chip provides high quality rendering and supports numerous 3D features including: bilinear and trilinear flitering, superior alpha blending, specular highlighting, fogging and anti-aliasing. The MGA-G200 also supports multiple texture rendering required to produce stunning lighting and atmospheric effects in games such as Quake II from id Software.
The MGA-G200 features a fully programmable, ultra-pipelined floating point setup and culling engine. The chip's setup engine is capable of calculating full-featured Direct3D and OpenGL triangles, strips, fans and vectors. Additionally, the setup engine moves up the 3D pipeline to increase performance by including inline backface culling. The setup engine combines parallel execution of multiple instructions along with an ultra-pipelined command flow to process triangle data resulting in over 1.5 million triangle per second rendering capability. The MGA-G200 was designed to maximize the parallel operation of the triangle setup engine and the fill engine. By separating these two engines in the MGA-G200, the triangle fill rate can remain high even for small triangles. This is a noticeable benefit when comparing the real world performance of 3D applications between different engines.
One of the cornerstones of the new 3D architecture of the MGA-G200 is Vibrant Color Quality rendering (VCQ). VCQ rendering uses 32-bpp color accuracy throughout the rendering pipeline and is capable of rendering displayed images in 32-bpp color from source texture maps of up to 32-bpp color. Even when the display is set to 16-bpp rendering, all internal calculations are executed with 32-bit accuracy and the end result is dithered down to 16-bpp from a true color palette. Finally, the VCQ architecture provides high quality analog color output to deliver sharp contrast and vibrant colors to avoid bland or washed out images.
The MGA-G200 supports high precision 32-bpp Z-buffering. Long considered a requirement in the CAD market, the exacting depth precision and enhanced image quality provided by 32-bit Z-buffering will also become a requirement in next generation gaming applications. With 32-bit Z-buffering enabled, the MGA-G200 is capable of rendering 3D scenes at a maximum resolution of up to 1280 x 1024 at 32-bpp color, double-buffered. The MGA-G200 also leverages its large frame buffers to support triple buffering to increase the frame rate of 3D applications without experiencing any tearing artifacts.
The MGA-G200 employs a new Symmetric Rendering Architecture (SRA) which takes full advantage of new performance opportunities made available by the presence of a high speed AGP 2x bus. Within the SRA, the MGA-G200 treats AGP memory exactly as if it were local video memory. This means that in a fully bus mastered fashion, the MGA-G200 can draw to, render to and read from AGP memory. By using the SRA as a key component of 3D, 2D and video operations, the MGA-G200 can benefit from the 528MB/sec of bandwidth provided by the AGP 2x bus in parallel with the extremely fast local memory to achieve the highest level of performance possible in all application areas.
The MGA-G200 leverages the SRA to implement a hierarchical texturing system for 3D rendering. With this architecture, the MGA-G200 offers the storage of surfaces - either textures or bitmaps - across three levels. The first level is the large on-chip cache, the second level is the up to 16MB of local video memory and the third level is the AGP memory. The SRA is designed to take advantage of a local video memory pool of up to16MB so users can benefit from features such as high resolution triple buffering and extra texture storage to provide higher performance. The SRA also helps accelerate 2D and video operations by permitting fast blits of bit maps between frame buffer and system memory and by enabling all drawing operations directly in system memory. The effective amount of "off screen" memory available to the drawing engine is greatly increased.
This large increase in "off screen" memory provides a larger cache for bitmap storage, the building of construction buffers and other functions. Windows also requires certain 2D drawing functions to occur in system memory and these functions are all accelerated with the Matrox SRA.
With more video in business presentations, commercial displays and on the web, high quality video acceleration has become increasingly important. The MGA-G200 has a high quality, high performance video scaling engine which incorporates both a front-end and a back-end scaler to produce full screen, full motion video.
The MGA-G200's video engine supports independent X and Y scaling with high quality filtering for top image quality even when running at full screen. It also supports any number of video windows and sprites simultaneously, a feature which is especially useful for tasks like security monitoring. The MGA-G200 architecture also enables the use of video streams as texture maps to increase the degree of realism in 3D applications.
Multimedia connectivity is a key feature for a powerful graphics solution in today's market. Growing from Matrox's success with comprehensive video upgrade options, the MGA-G200 features a video input port, video CODEC port and video output port. Providing high performance interfaces for external Video Decoder, Video Encoder, hardware Motion-JPEG and hardware MPEG-2 CODECs makes the MGA-G200 the centerpiece of a fully multimedia upgradeable solution.
A Montreal-based graphic chip designer and board manufacturer, Matrox Graphics Inc. is a leader in the graphics industry. In 1993, Matrox was the first to introduce a 64-bit graphics engine for the PC market with its MGA Series and the first to introduce high quality video editing to the consumer market with its breakthrough Rainbow Runner Series in 1997. Matrox's innovative graphics and video technology has been recognized with an unprecedented 300 international awards over the last five years. With 20 years of experience, Matrox continues to deliver leading edge solutions to the graphics and video industry at the best price/performance.