•1 min read•from Machine Learning
[R] Designing AI Chip Software and Hardware
![[R] Designing AI Chip Software and Hardware](/_next/image?url=https%3A%2F%2Fexternal-preview.redd.it%2FxsGciKHQ0vtQ9puYu8PX9xKM_1BsKKW1bfoZAS1me8k.png%3Fwidth%3D640%26crop%3Dsmart%26auto%3Dwebp%26s%3D6efb148062d666b80a9d5a6974aa6cdc60ff8746&w=3840&q=75)
| This is a detailed document on how to design an AI chip, both software and hardware. I used to work at Google on TPUs and at Nvidia on GPUs, so I have some idea about this, though the design I suggest is not the same as TPUs or GPUs. I also included many anecdotes from my career in Silicon Valley. Background This doc came to be because I was considering making an AI hw startup and this was to be my plan. I decided against it for personal reasons. So if you're running an AI hardware company, here's what a competitor that you now won't have would have planned to do. Usually such plans would be all hush-hush, but since I never started the company, you can get to know about it. [link] [comments] |
Want to read more?
Check out the full article on the original site