The development of the modern graphics processor begins with the introduction of the first 3D add-on cards in 1995, followed by the widespread introduction of the 32-bit operating systems and the affordable personal computer.
The pre-existing graphics industry largely consisted of a more prosaic, PC-less 2D architecture, with graphics cards better known for their chip's alphanumeric naming conventions and their huge price tags. 3D Games and Virtualization PC graphics were eventually merged from sources as diverse as arcade and console games, military, robotics and space simulators, and medical imaging.
The beginnings of 3D consumer graphics were a wild west of competing ideas. From the implementation of the hardware to the use of various rendering techniques and their application and data interfaces to permanent name exaggeration. The early graphics systems had a Fixed Function Pipeline (FFP) and architecture that followed a very rigid processing path and used almost as many graphics APIs as there were 3-D chip manufacturers.
While 3D graphics turned a fairly boring PC industry into a light and magic show, it owes its existence to generations of innovative efforts. This is the first in a series of four articles that, in chronological order, provide a comprehensive look at the history of the GPU. From the beginnings of 3D consumer graphics to the 3Dfx Voodoo Game Changer to the consolidation of the industry around the turn of the century and today's modern GPGPU.
1976 – 1995: The beginnings of 3D consumer graphics
The first true 3D graphics started with early display controllers known as video shifters and video address generators. They acted as a passage between the main processor and the display. The incoming data stream has been converted to serial bitmap video output such as luminance, color, and vertical and horizontal composite sync, keeping the line of pixels in a display generation and syncing each successive line along with the blanking interval (the time between ending a scan line and starting the next).
In the second half of the 1970s, a deluge of designs came to the fore for 3D graphics as we know it. For example, the RCA video chip "Pixie" (CDP1861) was able to output an NTSC-compatible video signal with a resolution of 62 x 128 or 64 x 32 for the unfortunate RCA Studio II console in 1976.
The video chip was quickly followed a year later by the Television Interface Adapter (TIA) 1A, which was integrated into the Atari 2600 to generate the screen display, sound effects and read input controls. The development of the TIA was led by Jay Miner, who later also led the design of the customer-specific chips for the Commodore Amiga computer.
In 1978 Motorola introduced the MC6845 video address generator. This became the basis for the 1981 IBM PC's Moch / CDA (Monochrome and Color Display Adapter) cards, and provided the same functionality for the Apple II. Motorola added the MC6847 video display generator later that year, which made its way found in a number of first generation PCs, including the Tandy TRS-80.
A similar solution from Commodore's MOS Tech subsidiary, the VIC, provided graphics output for Commodore home computers from 1980-83.
In November of the following year, LSI's ANTIC (Alphanumeric Television Interface Controller) and CTIA / GTIA (Color or Graphics Television Interface Adapter) co-processor made their debut in the Atari 400. ANTIC processed 2D display instructions with direct memory access (DMA). Like most video co-processors, it could produce field graphics (background, title screens, scoring display) while the CTIA produces colors and moving objects. Yamaha and Texas Instruments supplied similar ICs to a variety of previous home computer suppliers.
The next steps in graphic development were mostly in the professional field.
Intel used its 82720 graphics chip as the basis for the $ 1000 iSBX 275 Video Graphics Controller Multimode Board. It was able to display eight color data with a resolution of 256 x 256 (or monochrome with 512 x 512). The 32 KB display memory was sufficient to draw lines, arcs, circles, rectangles and character bitmaps. The chip was also designed for zooming, dividing the screen, and scrolling.
SGI quickly deployed its IRIS graphics for workstations – a GR1.x graphics card with separate add-in cards (daughter cards) for color options, geometry, Z-buffer and overlay / underlay.
Intel's $ 1000 iSBX 275 Video Graphics Controller multimode board could display eight color data at 256 x 256 (or monochrome 512 x 512) resolution.
Industrial and military 3D virtualization was relatively well developed at the time. IBM, General Electric, and Martin Marietta (who were supposed to buy GE's aerospace division in 1992) worked with a number of military contractors, technology institutes, and NASA on various projects that required the military and space simulation technology. In 1951, the Navy also developed a flight simulator with 3D virtualization from MIT's Whirlwind computer.
In addition to defense companies, there were companies that found themselves in the military markets with professional graphics.
<iframe allowfullscreen = "" frameborder = "0" height = "450" src = "https://www.youtube.com/embed/06mbwNg1Vw4?start=60&autoplay=1" width = "600" load = "lazy "srcdoc ="
Evans & Sutherland, who were supposed to offer professional line graphics cards like Freedom and REALimage, also provided graphics for the CT5 flight simulator, a $ 20 million package powered by a DEC PDP-11 mainframe. Ivan Sutherland, co-founder of the company, developed a computer program called Sketchpad in 1961 that could draw geometric shapes and display them on a CRT in real time with a light pen.
This was the forerunner of the modern graphical user interface (GUI).
In the less esoteric field of personal computing, the EGA (Extended Graphics Adapter) of the 82C43x series from Chips and Technologies offered the IBM adapters much-needed competition and was installed in many PC / AT clones by 1985. The year was also remarkable for the Commodore Amiga, which came with the OCS chipset. The chipset consisted of three main component chips – Agnus, Denise, and Paula – which made it possible to make a certain amount of graphics and audio computations independent of the CPU.
In August 1985, three Hong Kong immigrants, Kwok Yuan Ho, Lee Lau, and Benny Lau, founded Array Technology Inc in Canada. By the end of the year, the name was changed to ATI Technologies Inc.
ATI released its first product the following year, the OEM Color Emulation Card. It was used to output monochrome green, amber, or white fluorescent text against a black background via a 9-pin DE-9 connector on a TTL monitor. The card had at least 16K of memory and was responsible for a large percentage of ATI's sales of $ 10 million in its first year of operation. This was largely done through a contract that supplied Commodore Computers with around 7000 chips a week.
ATI's color emulation card had at least 16 KB of memory and was responsible for a large proportion of the company's sales of $ 10 million in its first year of operation.
The advent of color monitors and the lack of a standard among competitors eventually led to the formation of the Video Electronics Standards Association (VESA), of which ATI was a founding member along with NEC and six other graphics adapter manufacturers.
In 1987 ATI expanded its product line for OEMs to include the Graphics Solution Plus series, which used the 8-bit PC / XT ISA bus from IBM for IBM PCs with Intel 8086/8088. The chip supported MDA, CGA and EGA graphic modes via dip switches. It was basically a clone of the Plantronics Colorplus board, but with room for 64K of memory. Paradise Systems' PEGA1, 1a and 2a (256 kB) published in 1987 were also Plantronics clones.
EGA Wonder series 1 through 4 arrived in March for $ 399 and offer 256 KB of DRAM and compatibility with CGA, EGA, and MDA emulation with up to 640 x 350 and 16 colors. Extended EGA was available for series 2,3 and 4.
The high-end was rounded off by the EGA Wonder 800 with 16-color VGA emulation and support for a resolution of 800 x 600 as well as the VIP card (VGA Improved Performance), which is basically an EGA Wonder a digital-to-analog DAC (DAC) traded limited VGA compatibility. The latter was $ 449 plus $ 99 for the Compaq expansion module.
ATI was far from being alone and tackling the wave of consumer appetites for personal computing.
Many new companies and products were added this year. These included Trident, SiS, Tamerack, Realtek, Oak Technology, G-2 Inc. by LSI, Hualon, Cornerstone Imaging and Winbond – all founded in 1986-87. In the meantime, companies such as AMD, Western Digital / Paradise Systems, Intergraph, Cirrus Logic, Texas Instruments, Gemini, and Genoa would produce their first graphics products during this period.
ATI's Wonder series has been brought up to date over the next several years.
In 1988, the Small Wonder graphics solution with game controller port and composite-out options (for CGA and MDA emulation) as well as the EGA Wonder 480 and 800+ with extended EGA and 16-bit VGA support as well as the VGA Version available Wonder and Wonder 16 with additional VGA and SVGA support.
A Wonder 16 was equipped with 256 KB of memory that retailed for $ 499, while a 512 KB variant was $ 699.
An updated VGA Wonder / Wonder 16 series arrived in 1989, including the low-cost VGA Edge 16 (Wonder 1024 series). The new functions included a bus mouse connection and support for the VESA Feature Connector. This was a Goldfinger connector that resembled a shortened data bus slot connector and was connected to another video controller with a ribbon cable to bypass an overloaded data bus.
The Wonder series updates were further accelerated in 1991. The Wonder XL card added VESA 32K color compatibility and a Sierra RAMDAC, increasing the maximum screen resolution to 640 x 480 at 72 Hz or 800 x 600 at 60 Hz. Prices ranged from $ 249 (256K), $ 349 (512K), and $ 399 for the 1MB RAM option. A lower cost version called the VGA Charger, based on the previous year's Basic-16, was also made available.
The Mach series started in May of this year with the Mach8. It was sold either as a chip or as a circuit board that made it possible to outsource limited 2D drawing processes such as line drawing, color filling and bitmap combination (bit BLIT) via a programming interface (AI). ATI added a variation of the Wonder XL with a Creative Sound Blaster 1.5 chip on an expanded circuit board. Known as VGA Stereo-F / X, it was able to simulate stereo from Mono Sound Blaster files at approximate FM radio quality.
Graphics cards like the ATI VGAWonder GT offered a 2D + 3D option, in which the Mach8 was combined with the graphics core (28800-2) of the VGA Wonder + for its 3D tasks. Wonder and Mach8 led ATI through the $ 100 million sales milestone for the year, largely driven by the introduction of Windows 3.0 and the increased 2D workloads that came with it.
S3 Graphics was founded in early 1989 and eighteen months later produced its first 2D accelerator chip and graphics card, the S3 911 (or 86C911). The main specifications for the latter included 1MB of VRAM and 16-bit color support.
The S3 911 was superseded by the 924 in the same year – it was basically a revised 911 with 24-bit color – and again updated the following year with the 928 with 32-bit color and the 801 and 805 accelerators. The 801 used an ISA interface while the 805 used VLB. Between the introduction of the 911 and the advent of the 3D accelerator, the market was flooded with 2D GUI designs based on the original from S3 – specifically from Tseng Labs, Cirrus Logic, Trident, IIT, Mach32 from ATI, and MAGIC RGB from Matrox.
In January 1992, Silicon Graphics Inc (SGI) released OpenGL 1.0, a vendor-independent API (application programming interface) for 2D and 3D graphics.
Microsoft developed its own competing API called Direct3D and wasn't exactly thrilled with making sure OpenGL ran as well as possible on Windows.
OpenGL was developed from SGI's proprietary API called IRIS GL (Integrated Raster Imaging System Graphical Library). It was an initiative to keep IRIS non-graphical functions away and run the API on non-SGI systems as competing vendors emerged with their own proprietary APIs.
Originally geared towards the professional UNIX-based markets, OpenGL was quickly adopted for 3D games with developer-friendly support for implementing extensions.
Microsoft developed its own competing API called Direct3D and wasn't exactly keen on making sure OpenGL would run as well as possible on the new Windows operating systems.
It came to a head a few years later when id Software's John Carmack, whose previously published Doom had revolutionized PC gaming, ported Quake for using OpenGL on Windows and openly criticized Direct3D.
Microsoft's intransigence increased when they refused to license OpenGL's mini-client driver (MCD) on Windows 95, which allowed vendors to choose which features would have access to hardware acceleration. SGI responded by developing the Installable Client Driver (ICD) which not only did the same capability but made it even better as MCD only covered rasterization and ICD added lighting and transformation (T&L) functionality.
During the rise of OpenGL, which first gained a foothold in the workstation space, Microsoft was busy watching the emerging game market with designs on its own proprietary API. They acquired RenderMorphics in February 1995, the Reality Lab API of which became popular with developers and became the core of Direct3D.
Around the same time, 3dfx's Brian Hook wrote the Glide API, which would become the dominant API for games. This was partly due to Microsoft's involvement in the Talisman project (a tile-based rendering ecosystem) that diluted the resources dedicated to DirectX.
When D3D became generally available due to the introduction of Windows, proprietary APIs such as S3d (S3), Matrox Simple Interface, Creative Graphics Library, C Interface (ATI), SGL (PowerVR), NVLIB (Nvidia), RRedline (Rendition), and Glide was starting to lose favor with developers.
It didn't help that some of these proprietary APIs were allied with board manufacturers who were under increasing pressure to add to a rapidly growing feature list. This included higher screen resolutions, increased color depth (from 16 bits to 24 and then 32), and image quality improvements such as anti-aliasing. All of these features required higher bandwidth, graphics efficiency, and faster product cycles.
By 1993, market volatility had already forced a number of graphics companies to go out of business or to be absorbed by competitors.
1993 brought a number of new graphics competitors, notably Nvidia, founded by Jen-Hsun Huang, Curtis Priem, and Chris Malachowsky in January of that year. Huang was previously Director of Coreware at LSI, while Priem and Malachowsky both came from Sun Microsystems, where they had previously developed the SunSPARC-based GX graphics architecture.
The newcomers Dynamic Pictures, ARK Logic and Rendition joined Nvidia shortly afterwards.
Market volatility had already forced a number of graphics companies to go out of business or to be absorbed by competitors. Among them were Tamerack, Gemini Technology, Genua Systems, Hualon, Headland Technology (bought by SPEA), Acer, Motorola and Acumos (bought by Cirrus Logic).
One company that grew stronger, however, was ATI.
As the forerunner of the All-In-Wonder series, the 68890 PC TV decoder chip from ATI was announced at the end of November, which is included in the Video-It! Map. The chip was able to record videos at 320 x 240 @ 15 fps or 160 x 120 @ 30 fps and, thanks to the integrated Intel i750PD VCP (Video Compression Processor), compress / decompress in real time. It was also able to communicate with the graphics card over the data bus, eliminating the need for dongles or ports and ribbon cables.
The Video-It! retails for $ 399, while a lesser-known model called the Video-Basic completed the offer.
Five months later, in March, ATI belatedly introduced a 64-bit accelerator. the Mach64.
The fiscal year had not been good for ATI with a loss of $ 2.7 million as it slipped in the face of fierce competition in the market. Competing boards included the S3 Vision 968, which has been adopted by many board vendors, and the Trio64, which has OEM contracts from Dell (Dimension XPS), Compaq (Presario 7170/7180), AT&T (Globalyst), and HP (Vectra VE 4) and DEC (Venturis / Celebris).
The Mach64, released in 1995, achieved a number of remarkable innovations. It was the first graphics adapter available for PC and Mac computers in the form of the Xclaim ($ 450 and $ 650 depending on built-in storage) and, along with the trio of S3, offered acceleration of video playback in full motion.
The Mach64 also introduced ATI's first ever Pro graphics cards, the 3D Pro Turbo and 3D Pro Turbo + PC2TV, which were a cool $ 599 for the 2MB option and $ 899 for the 4MB option costs.
The following month, a technology start-up called 3DLabs rose, born when DuPont's pixel graphics division bought the subsidiary from its parent company, along with the GLINT 300SX processor, which enables OpenGL rendering, fragment processing, and rasterization. Due to their high price, the company's cards were originally aimed at the professional market. The Fujitsu Sapphire2SX 4MB sold for $ 1,600-2,000 while an 8MB ELSA GLoria 8 retails for $ 2,600- $ 2,850. However, the 300SX was intended for the gaming market.
S3 seemed to be everywhere at the time. The high-end OEM was dominated by the company's Trio64 chipsets, which integrated DAC, a graphics controller, and a clock synthesizer into a single chip.
The 1995 gaming GLINT 300SX had a greatly reduced 2 MB memory. It used 1MB for textures and z-buffers and the other for frame buffers, but offered the option to add another $ 50 to the VRAM for Direct3D compatibility from the base price of $ 349. The card failed to make headway in an already crowded market, but 3DLabs was already working on a successor in the Permedia range.
S3 seemed to be everywhere at the time. The high-end OEM was dominated by the company's Trio64 chipsets, which integrated DAC, a graphics controller, and a clock synthesizer into a single chip. They also used a unified frame buffer and supported hardware video overlay (a dedicated portion of graphics memory for rendering video as the application requires). The Trio64 and its 32-bit memory bus sibling Trio32 were available as OEM units and standalone cards from vendors such as Diamond, ELSA, Sparkle, STB, Orchid, Hercules, and Number Nine. Diamond Multimedia's prices ranged from $ 169 for a ViRGE-based card to $ 569 for a Trio64 + -based Diamond Stealth64 video with 4MB of VRAM.
The mainstream end of the market also included offerings from Trident, a long-time OEM supplier of no-frills 2D graphics adapters, which recently added the 9680 chip to its range. The chip was packed with most of the Trio64's features, and the boards were generally priced between $ 170 and $ 200. They offered acceptable 3D performance in this mount with good video playback ability.
Other newcomers to the mainstream market were Weiteks Power Player 9130 and Alliance Semiconductors ProMotion 6410 (usually referred to as Alaris Matinee or FIS OptiViewPro). Both offered excellent scaling at CPU speed, while the latter combined the powerful scaling engine with anti-blocking circuitry to achieve smooth video playback that was much better than previous chips like the ATI Mach64, Matrox MGA 2064W, and S3 Vision968.
Nvidia released its first graphics chip, the NV1, in May and was the first commercial graphics processor to support 3D rendering, video acceleration, and built-in GUI acceleration.
They have teamed up with ST Microelectronic to manufacture the chip in their 500 nm process. The latter also advertised the STG2000 version of the chip. While it wasn't a huge success, it was the company's first financial return. Unfortunately for Nvidia, Microsoft finalized and released DirectX 1.0 when the first vendor boards (especially Diamond Edge 3D) shipped in September.
The D3D graphics API confirmed that it relied on rendering triangular polygons that the NV1 used quad texture mapping on. Limited D3D compatibility was added through the driver to package triangles as square faces, but a lack of games tailored for the NV1 made the card an all-rounder, master of nothing.
Most of the games were ported by the Sega Saturn. A 4MB NV1 with integrated Saturn connectors (two per expansion bracket connected to the card with a ribbon cable) sold for around $ 450 in September 1995.
Due to late changes by Microsoft and the introduction of the DirectX SDK, board card manufacturers could not directly access hardware for digital video playback. This meant that virtually all discrete graphics cards had functional problems under Windows 95. Drivers under Win 3.1 from various companies, however, were generally free of errors.
The first public demonstration was at the E3 video game conference in Los Angeles in May of the following year. The card itself became available a month later. The 3D rage fused the Mach64's 2D core with its 3D capability. ATI announced its first 3D accelerator chip, the 3D Rage (also known as the Mach 64 GT) in November 1995.
Late revisions to the DirectX specification resulted in 3D Rage having compatibility issues with many games that used the API – mainly the lack of deep buffering. With an integrated 2 MB EDO RAM frame buffer, the 3D modality was limited to 640 x 480 x 16 bits or 400 x 300 x 32 bits. Attempting 32-bit color at 600 x 480 generally resulted in color damage to the screen, and 2D resolution peaked at 1280 x 1024. If gaming performance was mediocre, MPEG playback in full screen mode would have at least one made a certain contribution to balancing the scope of functions.
The performance race was over before the start and the 3Dfx Voodoo graphics effectively destroyed all competitions.
ATI revised the chip and in September the Rage II launched. In addition to supporting MPEG2 playback, the D3DX problems of the first chip were fixed. However, the first cards still shipped with 2MB of memory, which slowed performance and had issues with the perspective / geometry transformation. When the series was expanded to include the Rage II + DVD and 3D Xpression +, the storage capacity options grew to 8 MB.
While ATI was the first to bring a 3D graphics solution to market, it wasn't long before other competitors came out with different ideas about 3D implementation. Namely 3dfx, Rendition and VideoLogic.
<iframe allowfullscreen = "" frameborder = "0" height = "450" src = "https://www.youtube.com/embed/7EaVzIFm4iM?rel=0&autoplay=1" width = "600" load = "lazy "srcdoc ="
In the race to launch new products, 3Dfx Interactive Rendition and VideoLogic won. However, the performance race was over before the start and the 3Dfx Voodoo Graphics effectively destroyed all competitions.
This article is the first in a series of four. If you enjoyed this, read on as we take a stroll through the heyday of 3Dfx, Rendition, Matrox, and the fledgling company Nvidia.