Presentation
2019 IEEE Seymour Cray Computer Engineering Award
Presenter
Event Type
Awards Presentation
TP
TimeWednesday, 20 November 20198:30am - 9am
LocationMile High Ballroom
DescriptionOrigins of GPU Computing
GPU Computing didn’t just happen. The origin story of GPUs starts with VGA controllers and graphics accelerators. Texture mapping, multi-texture capabilities and frame buffer blending added a very small amount of programmability. Vertex and pixel shaders added real programming – a primitive programming model, but highly parallel processing for many vertices and pixels. The GPGPU movement recognized that GPUs were among the most powerful processors available, but very hard to program. For GPUs to be generally useful, high level programming languages and features were needed. Each step toward increasing capability required ideas, vision, and the hard work of many people who shared that vision. A modern GPU is a mainstream processor that is pervasive and ubiquitous in HPC, data centers, gaming, and many other applications.
GPU Computing didn’t just happen. The origin story of GPUs starts with VGA controllers and graphics accelerators. Texture mapping, multi-texture capabilities and frame buffer blending added a very small amount of programmability. Vertex and pixel shaders added real programming – a primitive programming model, but highly parallel processing for many vertices and pixels. The GPGPU movement recognized that GPUs were among the most powerful processors available, but very hard to program. For GPUs to be generally useful, high level programming languages and features were needed. Each step toward increasing capability required ideas, vision, and the hard work of many people who shared that vision. A modern GPU is a mainstream processor that is pervasive and ubiquitous in HPC, data centers, gaming, and many other applications.