Ten Lessons From Three Generations Shaped Google’s TPUv4i : Industrial Product
2021pp. 1–14
Citations Over TimeTop 1% of 2021 papers
Norman P. Jouppi, Doe Hyun Yoon, Matthew B. Ashcraft, Mark Gottscho, Thomas B. Jablin, George Thomas Kurian, James Laudon, Sheng Li, Peter Ma, Xiaoyu Ma, Thomas Norrie, Nishant Patil, Sushma Prasad, Cliff Young, Zongwei Zhou, David A. Patterson
Abstract
Google deployed several TPU generations since 2015, teaching us lessons that changed our views: semi-conductor technology advances unequally; compiler compatibility trumps binary compatibility, especially for VLIW domain-specific architectures (DSA); target total cost of ownership vs initial cost; support multi-tenancy; deep neural networks (DNN) grow 1.5X annually; DNN advances evolve workloads; some inference tasks require floating point; inference DSAs need air-cooling; apps limit latency, not batch size; and backwards ML compatibility helps deploy DNNs quickly. These lessons molded TPUv4i, an inference DSA deployed since 2020.
Related Papers
- → A Browser Compatibility Testing Method Based on Combinatorial Testing(2003)23 cited
- → Applying combinatorial method to test browser compatibility(2003)7 cited
- → Interoperability and Compatibility(2010)1 cited
- → High-Definition Television and Compatibility with Existing Standards(1983)19 cited
- Study on Browsers Compatibility of Cascading Style Sheets(2012)