How Did Retro Consoles Deal with Graphics Earlier than Trendy GPUs?

Console graphics got here from bespoke, fixed-function chips for greater than 20 years, till the unique Xbox burst onto the scene in 2001 with a PC GPU inside. Those of us who performed on early consoles just like the NES could have come throughout some acquainted quirks with these fixed-function chips: When too many Goombas lined up on the identical row in Super Mario Bros., for instance, a few of them would begin to flicker. That’s the NES’s graphics chip hitting its arduous restrict of eight sprites per scanline; the chip drops something previous eight, and the sport alternates which of them.

In the NES, it was the Ricoh 2C02 Picture Processing Unit that rendered recreation graphics. A fairly rudimentary chip, it could not run any code. Instead, it took tile paintings from the cartridge and a few lookup tables from reminiscence and turned them right into a TV sign, one scanline at a time. That association — a devoted image generator fed by the CPU — is how most consoles rendered their graphics from the late Nineteen Seventies into the 2000s.

When Microsoft’s first Xbox got here out in November 2001, it did not look very similar to any console earlier than it. Instead, it was constructed round an Intel Pentium III and a graphics chip NVIDIA made specifically for it, based mostly on the GeForce 3. The GeForce 3 had solely been introduced that February, and it was the primary GPU that permit builders program their very own pixel and vertex shaders.

The NES’ Ricoh 2C02 labored from about 10kB of reminiscence. That’s damaged down into roughly 8kB of tile shapes saved on the cartridge and 2kB of RAM within the console holding a background map or two. Separate small recollections contained in the chip tracked issues just like the place, form and orientation of as much as 64 sprites. Its output was 256×240 pixels, constructed from 8×8 tiles.

As the CRT TV’s electron beam swept throughout every line of the display screen, the 2C02 labored out which 8×8 tile belonged at every spot. It then cross-referenced this with which sprites have been sat on that line and determined which pixel to indicate. This came about in line with the beam, so the CPU could not safely rewrite the image whereas it was being drawn.

Unlike the NES, the sooner Atari 2600 made the CPU additionally deal with graphics. Because its Television Interface Adaptor had no body buffer, the processor needed to feed new values to the chip for each scanline drawn by the beam. Just 5 years earlier in 1972, the Magnavox Odyssey had no processor or reminiscence, interval, drawing its dots and contours with discrete diode-transistor logic. All of those machines have been designed round a CRT, which is partly why retro games look better on old CRT TVs.

Sony confirmed off the PS2’s chips in March 1999. The CPU was the Emotion Engine, a 128-bit design Sony constructed with Toshiba, and graphics went to a separate chip known as the Graphics Synthesizer. In the ultimate consoles, the 2 ran at about 295 MHz and 147 MHz.

The Graphics Synthesizer did not do any geometry work, although. By the time a triangle reached it, it had already been positioned throughout the body, and all of the chip needed to do was fill it in with colour and texture. The geometry work occurred again on the Emotion Engine, in two vector models. The CPU might give VU0 odd jobs, whereas VU1 principally ran the 3D transforms that turned a recreation’s fashions into on-screen triangles. Sony and Toshiba’s engineers stated a single vector unit might handle 85 million of these transforms a second. While no two consoles have been the identical, many early 3D consoles equally shared the graphical load throughout a number of chips.

Interestingly, Sony had used the time period GPU for the 1994 PlayStation’s Toshiba-designed graphics chip. That’s 5 years earlier than NVIDIA marketed the GeForce 256 because the global community’s first GPU. But the working definition of a contemporary GPU, a chip that runs small applications written by the sport, solely got here with GeForce 3 in early 2001.

NVIDIA and Microsoft signed their Xbox deal in March 2000, and 5 days later, Bill Gates took to the stage and introduced the console. It would use an Intel Pentium III-class processor, a customized NVIDIA graphics processor, 64MB of reminiscence shared by each and an 8GB arduous drive. The completed field was unveiled at CES 2001 after which went on sale that November.

But consoles did not make the swap to PC-style {hardware} unexpectedly. The Nintendo GameCube launched the identical month with a PowerPC processor from IBM, and each main console used one for the subsequent decade: the Xbox 360 dropped Intel for 3 PowerPC cores from IBM in 2005, the PS3’s Cell chip was constructed by Sony, Toshiba and IBM round a PowerPC core and the Wii and Wii U each ran up to date variations of the GameCube’s IBM chip.

The graphics aspect had already gone PC-style, with the 360’s ATI chip and the PS3’s NVIDIA one, however the processors hadn’t. That modified in 2013, when the PS4 moved to eight AMD processor cores and a graphics design taken from AMD’s Radeon PC playing cards. Today’s PS5 and Xbox Series X every run eight AMD Zen 2 cores alongside AMD RDNA 2 graphics on the identical chip because the CPU. Ultimately, designing graphics chips from scratch for consoles stopped making sense as soon as off-the-shelf GPU architectures match a console’s finances.