2006

Jensen Huang

b. 1963

GPU Computing for AI

As NVIDIA's co-founder and leader, championed GPU-accelerated computing and CUDA, providing parallel-processing infrastructure that made the complex mathematics of large neural networks increasingly practical.

Huang co-founded NVIDIA in 1993 with Chris Malachowsky and Curtis Priem. Under his leadership, the company unveiled CUDA in November 2006: a parallel-computing architecture and programming environment that allowed developers to use GPUs for general-purpose computation, not only graphics. CUDA was created by NVIDIA's engineering and research teams, including Ian Buck and John Nickolls, rather than by Huang alone.

His contribution to AI was enabling hardware, software infrastructure and organisational leadership, not an AI algorithm. GPUs can perform many mathematical operations simultaneously, making them well suited to neural-network training. In 2012, Alex Krizhevsky, Ilya Sutskever and Geoffrey Hinton trained AlexNet on NVIDIA GPUs; its ImageNet breakthrough demonstrated the value of combining their neural-network research with GPU computing and Fei-Fei Li's ImageNet dataset.

Dated contributions

  1. 2006

    GPU Computing for AI

    Jensen Huang & NVIDIA's engineering and research teams

    As NVIDIA's co-founder and leader, championed GPU-accelerated computing and CUDA, providing parallel-processing infrastructure that made the complex mathematics of large neural networks increasingly practical.

    Announcement · 2006 — NVIDIA Unveils CUDA—The GPU Computing Revolution Begins ↗

Reference

Original work & historical records

Wikipedia