gpu
Tags: computers
History of NVIDIA GPU’s
- Until 2006 GPU design was correlated to the logical stages of the rendering API
- DirectX pipeline made this untable
- Stream Multiprocessor replaces this with Tesla release
- Could run vertex, fragement, and geometry kernels without distinction
- Ideas of a “farm” with a master renderer
- CUDA introduced as a superset of C99
- Turing release gave dedicated cores for AI and ray tracing
- Very reminiscent of pre-2006