The chapter delves into the measurement standard of trillions of operations per second (TOPS) in AI inference performance, specifically focusing on the use of 8-bit integers across tech products like Apple and Qualcomm. It explores the relevance of int 8 precision to neural processing units (NPUs) and compares TOPS between different products to highlight their performance based on this standard.
Sponsored by:
- DeleteMe: DeleteMe makes it quick, easy and safe to remove your personal data online.
- Fastmail: Make email yours. Fast, private email that’s just for you.
Become a member for ATP Overtime, ad-free episodes, member specials, and our early-release, unedited “bootleg” feed!