I read Stephen Witt’s The Thinking Machine: Jensen Huang, Nvidia, and the World’s Most Coveted Microchip last week as I am interested in the history of GPUs (Graphics Processing Units). I did find it frustrating in that it is another great Silicon Valley man history, which is a tired trope really, but interesting in what it tacitly admits. The use of GPUs for High Performance Computing before Machine Learning was accidental.
I had sort of known that as I was on the edge of a CUDA center of excellence several years ago that received the latest cards every so often. The last one that I had to install needed a notch taking out of the card to fit it into the machine as it was so large, especially with the fan. I was out of that arena before the mining craze really took off. Anyhow …
I very much doubt that I am the intended audience but it was gratifying to read about a company still run by folk with engineering knowledge, rather than pure business. That written, there is a well trodden path of a good idea, some luck, and growing by buying/crushing other companies to become the core node in the hardware network along with the toolkit to use it.
This book sits on the shelf against the other large companies. I found it breathless in the telling and slightly hagiographical, but that could be the price of access. I am left wondering how the CUDA library potentially limits what can be done with it and how it will shape the humanities and cultural computing.