hardmaru: The Deep Learning Revolution and Its Implications for Computer Architecture and Chip Design
Paper from @JeffDean provides his views on specialized hardware for ML, how ML can play a role in chip design and manufacturing, and future ML research directions
https://arxiv.org/abs/1911.05289 https://t.co/IpGkWpLrQI
5 replies, 489 likes
Raziel Alvarez: Anybody attending ISSCC 2020 (http://isscc.org) next Sunday I'll be giving a talk on deep learning, #TensorFlow and Edge devices.
Also recommend attending @JeffDean's talk on the implications of deep learning on chip design (http://arxiv.org/abs/1911.05289)
1 replies, 50 likes
OGAWA, Tadashi: =>
Domain Specific Architectures for DNNs: Three Generations of TPUs, David Patterson, Distinguished Lecture, UW, Oct 29, 2019 https://www.youtube.com/watch?v=VCScWh966u4
TPUv2 (46 TF (16b), TDP: 280 W, <611 mm^2, >12 nm)
4x ICI link: 500 Gb/s both-dir, 13%
TPUv3 (123 TF (16b), 40 W, <648, >12) https://t.co/YX119CBsAp
1 replies, 36 likes
RISC-V.chip.haus: “Google teaches AI to play the game of chip design”
📰 https://www.nextplatform.com/2020/02/20/google-teaches-ai-to-play-the-game-of-chip-design/
Cf. Jeff Dean’s ISSCC 2020 keynote presentation “The Deep Learning Revolution and Its Implications for Computer Architecture and Chip Design”. (Video recording not yet posted.) https://t.co/5bHGOIvAd4
1 replies, 15 likes
Hacker News: Deep Learning and Its Implications for Computer Architecture and Chip Design https://arxiv.org/abs/1911.05289
0 replies, 11 likes
Stephen Pimentel: The Deep Learning Revolution and Its Implications for Computer Architecture and Chip Design
by @JeffDean
https://arxiv.org/abs/1911.05289
0 replies, 11 likes
Chaitanya Joshi: Transformers can also handle images/graphs/sets/most other input representations. Are we closing in on @JeffDean's dream of massive and sparsely activated NNs that do everything?
https://arxiv.org/abs/1911.05289 https://t.co/u4lHXoI1HH
1 replies, 10 likes
LightOn: We love it! Our own co-design approach also includes the exciting electro-optical technology stack. https://t.co/dNg3rFUtxX
0 replies, 10 likes
Dave Gershgorn: Google's Jeff Dean essentially posted a blog on ArXiv the other day, talking about Google's investments in custom silicon.
"The Deep Learning Revolution and Its Implications for Computer Architecture and Chip Design"
https://arxiv.org/pdf/1911.05289.pdf
0 replies, 8 likes
OGAWA, Tadashi: =>
"The Deep Learning Revolution and Its Implications for Computer Architecture and Chip Design", Jeffrey Dean, Google, Nov 13, 2019 https://arxiv.org/abs/1911.05289
Companion paper to a keynote at the ISSCC 2020.
https://twitter.com/ogawa_tter/status/1187541384212836353
David Patterson https://twitter.com/ogawa_tter/status/1192329877963149312
TPUv2 / v3 https://t.co/ytyNeZ8HSJ
1 replies, 7 likes
OGAWA, Tadashi: "Guess which open source core is Google using to benchmarks their AI-driven IC placement engine?",
Prof. Luca Benini, Feb 18, 2020 https://twitter.com/LucaBeniniZhFe/status/1229506194991190016
Jeffrey Dean, Companion paper to a keynote at the ISSCC 2020
0 replies, 6 likes
Fei Peng: [1911.05289] The Deep Learning Revolution and Its Implications for Computer Architecture and Chip Design https://arxiv.org/abs/1911.05289
0 replies, 3 likes
OGAWA, Tadashi: =>
ISSCC 2020, Feb 16-20, 2020 http://isscc.org/
ISSCC 2020 Advanced Program http://submissions2.mirasmart.com/ISSCC2020/PDF/ISSCC2020AdvanceProgram.pdf
Zen 2, AMD Chiplet
IBM z15: A 12-Core 5.2GHz
ML: Alibaba
3GHz ARM Neoverse N1
Compute-In-Memory
Companion paper to a keynote, ISSCC 2020, Jeff Dean https://twitter.com/ogawa_tter/status/1197243295400726528 https://t.co/AdA7wXjJcX
1 replies, 3 likes
Muzaffer Kal: The Deep Learning Revolution and Its Implications for Computer Architecture and Chip Design
https://arxiv.org/abs/1911.05289
0 replies, 2 likes
Alessandro Baffa: "The Deep Learning Revolution and Its Implications for Computer Architecture and Chip Design" by Jeffrey Dean.
https://arxiv.org/abs/1911.05289
0 replies, 1 likes
HotComputerScience: Most popular computer science paper of the day:
"The Deep Learning Revolution and Its Implications for Computer Architecture and Chip Design"
https://hotcomputerscience.com/paper/the-deep-learning-revolution-and-its-implications-for-computer-architecture-and-chip-design https://twitter.com/hardmaru/status/1197020514671218688
0 replies, 1 likes
Milos Gajdos: Jeff Dean wrote a paper about the implications of Deep Learning on Computer architecture and chip design
https://arxiv.org/abs/1911.05289
0 replies, 1 likes
jafbm🧢: machine learning may be able to help with some aspects of the circuit design process #AI
https://arxiv.org/abs/1911.05289
0 replies, 1 likes
Found on Nov 20 2019 at https://arxiv.org/pdf/1911.05289.pdf