Parts/Accelerators/NVIDIA Tesla V100
Accelerators · Volta · 2017
NVIDIA Tesla V100
16/32 GB HBM2 with the first-generation tensor cores. Six per node in the Summit and Sierra AC922 design.
Why this part
Volta introduced tensor cores: dedicated units for small mixed-precision matrix multiplies, put there for deep learning and largely irrelevant to the simulation codes Summit and Sierra were bought to run. They turned out to matter enormously anyway, both because parts of scientific computing found uses for reduced precision and because the machines became the training grounds for methods that later dominated. Summit and Sierra are the last flagship US systems built on IBM, and V100 with NVLink 2.0 to POWER9 is the configuration that made them coherent.
Systems using this part
- Systems
- 91
- Rmax underneath
- 529.0 PFlop/s
- Units recorded
- 77,092
65 of 91 edges disclose a quantity
| System | Role | Country | Units | System Rmax | Supplied by | Rank |
|---|---|---|---|---|---|---|
| PARAM Ananta 2022-05-30 | Accelerator | India | 20 | - | NVIDIA | - |
| PARAM Ganga 2022-03 | Accelerator | India | - | - | NVIDIA | - |
| PARAM Pravega 2022-02-03 | Accelerator | India | 80 | - | NVIDIA | - |
| Altair 2021-05 | Accelerator | Poland | 72 | 3.53 PFlop/s | NVIDIA | #85 → |
| Bridges-2 2021-03 | Accelerator | USA | 280 | - | NVIDIA | - |
| Ares 2021 | Accelerator | Poland | 72 | 2.34 PFlop/s | NVIDIA | - |
| Dragão 2021 | Accelerator | Brazil | - | 8.98 PFlop/s | NVIDIA | #46 → |
| ABCI 2.0 2021 | Accelerator | Japan | 4,352 | 22.21 PFlop/s | NVIDIA | #12 → |
| Expanse 2020-12 | Accelerator | USA | 208 | 2.48 PFlop/s | NVIDIA | #153 → |
| Kelvin2 2020-10 | Accelerator | United Kingdom | 32 | - | NVIDIA | - |
| SCINet Atlas 2020-09-30 | Accelerator | USA | 8 | - | NVIDIA | - |
| Marconi-100 2020-05 | Accelerator | Italy | 3,920 | 21.64 PFlop/s | NVIDIA | #9 → |
| Topaz 2020-03 | Accelerator | Australia | 44 | - | NVIDIA | - |
| Gadi 2020 | Accelerator | Australia | - | 9.26 PFlop/s | NVIDIA | #24 → |
| HPC5 2020 | Accelerator | Italy | 7,280 | 35.45 PFlop/s | NVIDIA | #53 → |
| Dammam-7 2020 | Accelerator | Saudi Arabia | - | 22.40 PFlop/s | NVIDIA | #10 → |
| Phoenix (PACE) 2020 | Accelerator | USA | 148 | 1.84 PFlop/s | NVIDIA | #277 → |
| Atlas (Petrobras) 2020 | Accelerator | Brazil | - | 4.38 PFlop/s | NVIDIA | #56 → |
| JADE2 2020 | Accelerator | United Kingdom | 504 | 1.47 PFlop/s | NVIDIA | #16 → |
| Christofari 2019-12 | Accelerator | Russia | - | 6.67 PFlop/s | NVIDIA | #29 → |
| TX-GAIA 2019-09 | Accelerator | USA | 896 | 5.16 PFlop/s | NVIDIA | #42 → |
| MN-2 2019-07 | Accelerator | Japan | 1,024 | - | NVIDIA | - |
| NVIDIA DGX SuperPOD (V100) 2019-06 | Accelerator | USA | 1,536 | 9.44 PFlop/s | NVIDIA | #22 → |
| Béluga 2019-04 | Accelerator | Canada | 688 | 2.28 PFlop/s | NVIDIA | #106 → |
| Pangea III 2019 | Accelerator | France | - | 17.86 PFlop/s | NVIDIA | #61 → |
| Cygnus 2019 | Accelerator | Japan | 320 | 1.58 PFlop/s | NVIDIA | #264 → |
| Jean Zay V100 2019 | Accelerator | France | 1,044 | 4.48 PFlop/s | NVIDIA | #64 → |
| Cannon (FASRC) 2019 | Accelerator | USA | 64 | 2.08 PFlop/s | NVIDIA | #144 → |
| Fênix 2019 | Accelerator | Brazil | - | 3.16 PFlop/s | NVIDIA | #82 → |
| Ogbon 2019 | Accelerator | Brazil | - | 1.60 PFlop/s | NVIDIA | #347 → |
| Longhorn 2019 | Accelerator | USA | 448 | 2.29 PFlop/s | NVIDIA | #120 → |
| Satori (MIT) 2019 | Accelerator | USA | - | 1.46 PFlop/s | NVIDIA | #373 → |
| Puhti 2019 | Accelerator | Finland | - | 1.71 PFlop/s | NVIDIA | #280 → |
| Artemis (G42) 2019 | Accelerator | United Arab Emirates | - | 7.26 PFlop/s | NVIDIA | #19 → |
| Computerome 2.0 2019 | Accelerator | Denmark | 40 | - | NVIDIA | - |
| Barbora 2019 | Accelerator | Czechia | 32 | - | NVIDIA | - |
| cHARISMa 2019 | Accelerator | Russia | - | - | NVIDIA | - |
| MagicCube3 2019 | Accelerator | China | - | 2.11 PFlop/s | NVIDIA | #133 → |
| Hypercluster (Microsoft Research) 2019 | Accelerator | USA | - | 2.67 PFlop/s | NVIDIA | #95 → |
| Guacolda 2019 | Accelerator | Chile | 4 | - | NVIDIA | - |
| COBALT 2.0 2019 | Accelerator | Netherlands | 26 | - | NVIDIA | - |
| SWARM (Wuhan University) 2019 | Accelerator | China | 500 | 1.40 PFlop/s | NVIDIA | #479 → |
| Circe 2018-11 | Accelerator | USA | 576 | 3.06 PFlop/s | NVIDIA | #61 → |
| ABCI 2018-08 | Accelerator | Japan | - | 19.88 PFlop/s | NVIDIA | #7 → |
| Kay 2018-08 | Accelerator | Ireland | 64 | 665.00 TFlop/s | NVIDIA | - |
| Sierra 2018-06 | Accelerator | USA | 17,280 | 94.64 PFlop/s | NVIDIA | #62 → |
| Summit 2018-06 | Accelerator | USA | 27,648 | 148.6 PFlop/s | NVIDIA | #58 → |
| MareNostrum P9 CTE 2018-05 | Accelerator | Spain | 216 | 1.02 PFlop/s | NVIDIA | #9 → |
| Lassen 2018 | Accelerator | USA | 3,168 | 18.20 PFlop/s | NVIDIA | #10 → |
| JUWELS Cluster 2018 | Accelerator | Germany | 224 | 6.18 PFlop/s | NVIDIA | #171 → |
| COBRA 2018 | Accelerator | Germany | 128 | 5.61 PFlop/s | NVIDIA | #36 → |
| Govorun 2018 | Accelerator | Russia | 40 | - | NVIDIA | - |
| Zhores 2018 | Accelerator | Russia | 104 | - | NVIDIA | - |
| CLAIX-2018 2018 | Accelerator | Germany | 96 | 2.48 PFlop/s | NVIDIA | #92 → |
| Ansel (LLNL) 2018 | Accelerator | USA | 216 | 1.29 PFlop/s | NVIDIA | #471 → |
| RAIDEN 2017 | Accelerator | Japan | 432 | 1.21 PFlop/s | NVIDIA | #171 → |
| Iridis 5 2017 | Accelerator | United Kingdom | 20 | 860.00 TFlop/s | NVIDIA | #252 → |
| Raad2 2017 | Accelerator | Qatar | 8 | 120.00 TFlop/s | NVIDIA | - |
| DGX SaturnV 2016-11 | Accelerator | USA | - | 4.00 PFlop/s | NVIDIA | #29 → |
| Kebnekaise 2016-11 | Accelerator | Sweden | 10 | - | NVIDIA | - |
| Bridges 2016 | Accelerator | USA | 88 | - | NVIDIA | - |
| ARIS 2015-06 | Accelerator | Greece | 8 | 170.00 TFlop/s | NVIDIA | #468 → |
| Artemis (Sydney) 2015 | Accelerator | Australia | - | - | NVIDIA | - |
| Sherlock 2014-01 | Accelerator | USA | - | - | NVIDIA | - |
| GridKa 2002-10 | Accelerator | Germany | 56 | - | NVIDIA | - |
| Microsoft Azure AI supercomputer for OpenAI (2020) | Accelerator | USA | - | - | NVIDIA | - |
| Cyclone | Accelerator | Cyprus | 64 | - | NVIDIA | - |
| Rudens | Accelerator | Latvia | 8 | - | NVIDIA | - |
| farm22 | Accelerator | United Kingdom | 30 | - | NVIDIA | - |
| Rocket Cluster | Accelerator | Estonia | 44 | - | NVIDIA | - |
| HPC Saulėtekis | Accelerator | Lithuania | 32 | - | NVIDIA | - |
| Taiwania 2 | Accelerator | Taiwan | 2,016 | 9.00 PFlop/s | NVIDIA | #20 → |
| Great Lakes | Accelerator | USA | - | 860.00 TFlop/s | NVIDIA | - |
| Bede | Accelerator | United Kingdom | 128 | - | NVIDIA | - |
| Ibex | Accelerator | Saudi Arabia | - | - | NVIDIA | - |
| PARAM Shakti | Accelerator | India | 44 | - | NVIDIA | - |
| Alvis | Accelerator | Sweden | 44 | 1.94 PFlop/s | NVIDIA | #403 → |
| Triton | Accelerator | Finland | 288 | - | NVIDIA | - |
| DARWIN | Accelerator | USA | 12 | - | NVIDIA | - |
| Katana | Accelerator | Australia | - | - | NVIDIA | - |
| PARAM Sanganak | Accelerator | India | - | - | NVIDIA | - |
| Technion Zeus | Accelerator | Israel | 8 | - | NVIDIA | - |
| ISSAI Cluster | Accelerator | Kazakhstan | - | - | NVIDIA | - |
| Casper | Accelerator | USA | 64 | - | NVIDIA | - |
| Myriad | Accelerator | United Kingdom | - | - | NVIDIA | - |
| Hawk (Cardiff) | Accelerator | United Kingdom | 30 | - | NVIDIA | - |
| Sunbird | Accelerator | United Kingdom | 8 | - | NVIDIA | - |
| Hábrók | Accelerator | Netherlands | 24 | - | NVIDIA | - |
| Monash M3 | Accelerator | Australia | 56 | - | NVIDIA | - |
| PARAM Himalaya | Accelerator | India | 20 | - | NVIDIA | - |
| PARAM Yukti | Accelerator | India | 148 | - | NVIDIA | - |
“Units recorded” sums only the edges where a public source states a quantity. It is a floor, not a total, and should never be read as installed base.