Massive Acceleration through the Many-core Processor

Explore the potential of graphics processing units for massive hardware acceleration in this GOTO Aarhus 2012 conference talk. Delve into the historical evolution of programmable GPUs, examine real-world production code examples, and gain insights into the future of hardware-accelerated computing. Learn about the hidden computing power within graphics cards, their transition from real-time graphics to versatile acceleration tools, and the implications for various algorithms. Discover the challenges and opportunities of GPU exploitation, including pitfalls, graphics API limitations, and modern programming languages. Gain a comprehensive understanding of topics such as cascaded light propagation volumes, subsurface scattering, spatial data structures, and ray tracing. Evaluate whether GPU utilization is currently viable and when it may become essential in the future of heterogeneous processing units.

Introduction
A word of caution
Hardware accelerated graphics
GeForce 256
Highlevel languages
Lego Digital Designer
Cascaded light propagation volumes
Subsurface scattering
Multicore crisis
GPU exploitation
How I got into this
The pitfalls of TPU
Graphics API
Graphics Limitations
Modern Languages
Colonel
The Grid
Memory Layers
OpenCL vs CUDA
More Graphics
Spatial Data Structures
Interactive Structures
Molten Sequence
Ray Tracing