|
|
David Kirk was a born engineer, always inventing and building things. After Mechanical Engineering degrees at MIT (B.S., M.S.), David worked on CAD and workstations, and this led him to making pictures with computers. David obtained his Ph.D. at Caltech with a minor in Computation and Neural Systems, and specialized in computer graphics, VLSI, and analog computing. There was a strong synergy with these topics and neural networks, Analog VLSI, and neurobiology.
David drove graphics technology at video game developer Crystal Dynamics. This led to NVIDIA, building graphics chips for 3D graphics for PCs. As Chief Scientist and VP of Architecture, David led architecture and technology development for a series of products. In a few short years, single chip 3D PC graphics accelerators surpassed the performance of million dollar graphics systems built by SGI and others. During the course of this innovation, David was inventor or co-inventor of nearly 100 patents. Graphics Processing Units (GPUs) rapidly became the most powerful and versatile computing devices on the planet. As GPUs became more programmable, they presented a programming challenge; modern computer science had not contemplated devices with hundreds of processors and thousands of threads. The task of driving GPU adoption shifted from creating computer graphics to teaching parallel programming.
David founded the research division of NVIDIA, focusing on graphics architecture and algorithms, rendering applications and algorithms, and programming models.
David turned his attention to teaching parallel programming in computer science. David authored the hugely popular parallel programming textbook “Programming Massively Parallel Processors”. Now in its 3rd edition, this textbook sold 10s of 1000s of copies and was translated into more than 10 languages.
After the “AI Winter”, the GPU Big Bang showed that GPUs were an effective tool for making artificial neural networks practical on a large scale, in real world applications. It turns out that the massively parallel threaded architecture of GPUs is ideal for AI training and inference, more so even than many special purpose architectures. David continued his efforts of parallel programming education, now with a focus on AI. David continues to stay involved with education, teaching intensive master class programming summer school events, and remaining an Adjunct Professor at UIUC’s ECE department.
David drove graphics technology at video game developer Crystal Dynamics. This led to NVIDIA, building graphics chips for 3D graphics for PCs. As Chief Scientist and VP of Architecture, David led architecture and technology development for a series of products. In a few short years, single chip 3D PC graphics accelerators surpassed the performance of million dollar graphics systems built by SGI and others. During the course of this innovation, David was inventor or co-inventor of nearly 100 patents. Graphics Processing Units (GPUs) rapidly became the most powerful and versatile computing devices on the planet. As GPUs became more programmable, they presented a programming challenge; modern computer science had not contemplated devices with hundreds of processors and thousands of threads. The task of driving GPU adoption shifted from creating computer graphics to teaching parallel programming.
David founded the research division of NVIDIA, focusing on graphics architecture and algorithms, rendering applications and algorithms, and programming models.
David turned his attention to teaching parallel programming in computer science. David authored the hugely popular parallel programming textbook “Programming Massively Parallel Processors”. Now in its 3rd edition, this textbook sold 10s of 1000s of copies and was translated into more than 10 languages.
After the “AI Winter”, the GPU Big Bang showed that GPUs were an effective tool for making artificial neural networks practical on a large scale, in real world applications. It turns out that the massively parallel threaded architecture of GPUs is ideal for AI training and inference, more so even than many special purpose architectures. David continued his efforts of parallel programming education, now with a focus on AI. David continues to stay involved with education, teaching intensive master class programming summer school events, and remaining an Adjunct Professor at UIUC’s ECE department.