1 / 15

Workshop RENAFAE 2018 RENAFAE 10 ANOS https://indico.cern.ch/event/719214/ HPCs at LHCb

Workshop RENAFAE 2018 RENAFAE 10 ANOS https://indico.cern.ch/event/719214/ HPCs at LHCb. Renato Santana – São Paulo - 30/JUL/2018 Co-Authors: Stefan Roiser, Federico Stagni, Vladimir Romanoviskiy. Workshop RENAFAE 2018 HPCs at LHCb. Federico Stagni – 10th LHCb Comp Workshop

kiayada-blanchard
Télécharger la présentation

Workshop RENAFAE 2018 RENAFAE 10 ANOS https://indico.cern.ch/event/719214/ HPCs at LHCb

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Workshop RENAFAE 2018RENAFAE 10 ANOShttps://indico.cern.ch/event/719214/HPCs at LHCb Renato Santana – São Paulo - 30/JUL/2018 Co-Authors: Stefan Roiser, Federico Stagni, Vladimir Romanoviskiy

  2. Workshop RENAFAE 2018 HPCs at LHCb Federico Stagni – 10th LHCb Comp Workshop (https://indico.cern.ch/event/561982/ nov/17) Renato Santana – 30/JUL/2018

  3. Workshop RENAFAE 2018 HPCs at LHCb Federico Stagni – 10th LHCb Comp Workshop (https://indico.cern.ch/event/561982/) Renato Santana – 30/JUL/2018

  4. Workshop RENAFAE 2018 HPCs at LHCb Federico Stagni – 10th LHCb Comp Workshop (https://indico.cern.ch/event/561982/) Renato Santana – 30/JUL/2018

  5. Workshop RENAFAE 2018 HPCs at LHCb • Santos Dumont (SDumont) • Origin: French ATOS/BULL • Located in Petrópolis/Rio de Janeiro – Brazil • LHCb project for LHCb use, accepted: 2017 Renato Santana – 30/JUL/2018

  6. Workshop RENAFAE 2018 HPCs at LHCb The machine: It has 1.1 Petaflops/s installed capacity. The architecture is hybrid. Today SDumont has 18.144 CPUs on 756 nodes (24 core/node): . 504 nodes (thin node) each node with: .. 2 X CPU Intel Xeon E5-2695v2 Ivy Bridge, 2,4GHZ .. 24 core each. Total of 12.096 core. .. 64GB DDR3 RAM Renato Santana – 30/JUL/2018

  7. Workshop RENAFAE 2018 HPCs at LHCb The machine (cont.): . 198 nodes (thin nodes) with GPUs K40, each one has: .. 2 x CPU Intel Xeon E5-2695v2 Ivy Bridge, 2,4GHZ .. 24 core (12 per CPU), in total 4.752 core .. 64GB DDR3 RAM .. 2 x Nvidia K40 (GPU) . 54 nodes (thin node) with Xeon Phi, each one has: .. 2 x CPU Intel Xeon E5-2695v2 Ivy Bridge, 2,4GHZ .. 24 core (12 per CPU), total of 1.296 core .. 64GB DDR3 RAM .. 2 x Xeon PHI 7120 (dispositivo MIC) Renato Santana – 30/JUL/2018

  8. Workshop RENAFAE 2018 HPCs at LHCb The machine (cont): . 1 MESCA2 node with shared memory (fat node) with: .. 16 x CPU Intel Ivy, 2,4GHZ .. 240 core (15/CPU) .. 6TB RAM . All 756 nodes are linked through Infiniband FDR . File system LUSTRE with 1.7PB total and a secondary system with 640TB . Schedulling System:SLURM Renato Santana – 30/JUL/2018

  9. Workshop RENAFAE 2018 HPCs at LHCb The LHCb – DIRAC model: Ideal SDumont WN01 C V M F S 3.pilot Login Nodes SLURM WN02 • . • . • . 3.pilot 2.submission WN'n 1.pilots 4.Fetch job 5.job CERN 6.Job output LHCb DIRAC ... EOS Renato Santana – 30/JUL/2018

  10. Workshop RENAFAE 2018 HPCs at LHCb The LHCb – DIRAC model: 1st issue SDumont WN01 C V M F S Lustre FS 3.pilot Workernodes isolated from the world. Solved: jan/18 Login Nodes SLURM WN02 • . • . • . 3.pilot 2.submission WN'n 1.pilots 4.Fetch job 5.job CERN 6.Job output LHCb DIRAC ... EOS Renato Santana – 30/JUL/2018

  11. Workshop RENAFAE 2018 HPCs at LHCb The LHCb – DIRAC model: 2nd issue SDumont WN01 C V M F S Access to LoginNodes has to be through VPN 3.pilot Login Nodes SLURM WN02 • . • . • . 3.pilot 2.submission WN'n 1.pilots 4.Fetch job 5.job CERN 6.Job output LHCb DIRAC ... EOS Renato Santana – 30/JUL/2018

  12. Workshop RENAFAE 2018 HPCs at LHCb The LHCb – DIRAC model SDumont WN01 C V M F S Access to LoginNodes has to be through VPN: Solved 4/18 3.pilot Login Nodes SLURM WN02 • . • . • . 3.pilot 2.submission WN'n LHCb Access Host 4.Fetch job 1.pilots 5.job CERN 6.Job output LHCb DIRAC ... EOS Renato Santana – 30/JUL/2018

  13. Workshop RENAFAE 2018 HPCs at LHCb SDumont Production Renato Santana – 30/JUL/2018

  14. Workshop RENAFAE 2018 HPCs at LHCb What's next? Renato Santana – 30/JUL/2018

  15. Workshop RENAFAE 2018 HPCs at LHCb THANK YOU! Questions? Suggestions? Experiences? renato.santana@cern.ch Renato Santana – São Paulo - 30/JUL/2018 Co-Authors: Stefan Roiser, Federico Stagni, Vladimir Romanoviskiy

More Related