Hands-On GPU Computing with Python

Second generation

The first GPU was released by NVIDIA on August 31, 1999, known as GeForce 256. Capable of billions of calculations per second and processing 10 million polygons per second, it had 23 million transistors, 32 MB of 128-bit DRAM, 120 MHz core clock speed, and four 64-bit pipelines for rendering.

The accelerated graphics port (AGP) was first used by this generation of cards instead of the PCI bus, and the new graphics features they offered in hardware were multi-texturing, bump maps, light maps, and hardware geometry transform and lighting.

On this generation of cards, once the programmer sent graphics data into the GPU's pipeline, it could not be modified. So, such pipelines came to be known as fixed function pipelines. This ...