The Microarchitecture of DOJO, Tesla’s Exa-Scale Computer
IEEE Micro · 2023-05 · 43 (3) · 31-39
DOI: 10.1109/mm.2023.3258906 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Micro, 2023-05. Verfasst von Emil Talpes und 17 weiteren Autor:innen. Die Arbeit wurde bisher 53-mal zitiert (Crossref) und nennt 8 Literaturangaben.
Abstract (Originalsprache)
The Tesla-built DOJO system is a scalable solution targeted towards machine learning training applications. It is based on the D1 custom compute chip which packs together 354 independent processors, resulting in 362 TFLOPS of compute and 440 MB of internal static random-access memory storage. While maintaining full programmability, DOJO emphasizes distribution of resources and an extremely high bandwidth interconnect, allowing it to scale from small systems all the way to exaFLOP supercomputers.
Zitieren
APA
Talpes, E., Sarma, D. D., Williams, D., Arora, S., Kunjan, T., Floering, B., Jalote, A., Hsiong, C., Poorna, C., Samant, V., Sicilia, J., Nivarti, A. K., Ramachandran, R., Fischer, T., Herzberg, B., McGee, B., Venkataramanan, G., & Banon, P. (2023). The Microarchitecture of DOJO, Tesla’s Exa-Scale Computer. IEEE Micro, 43(3), 31-39. https://doi.org/10.1109/mm.2023.3258906
Vancouver
Talpes E, Sarma DD, Williams D, Arora S, Kunjan T, Floering B, et al. The Microarchitecture of DOJO, Tesla’s Exa-Scale Computer. IEEE Micro. 2023;43(3):31-39. doi:10.1109/mm.2023.3258906
BibTeX
@article{Talpes2023,
author = {Talpes, Emil and Sarma, Debjit Das and Williams, Doug and Arora, Sahil and Kunjan, Thomas and Floering, Benjamin and Jalote, Ankit and Hsiong, Christopher and Poorna, Chandrasekhar and Samant, Vaidehi and Sicilia, John and Nivarti, Anantha Kumar and Ramachandran, Raghuvir and Fischer, Tim and Herzberg, Ben and McGee, Bill and Venkataramanan, Ganesh and Banon, Pete},
title = {The Microarchitecture of DOJO, Tesla’s Exa-Scale Computer},
journal = {IEEE Micro},
year = {2023},
volume = {43},
number = {3},
pages = {31--39},
doi = {10.1109/mm.2023.3258906}
}
EUVIMED