Facilitating Electronic Structure Computations on GPU-based Exascale Platforms

Facilitating Electronic Structure Computations on GPU-based Exascale Platforms

Exascale Computing Project via YouTube Direct link

Acknowledgments

31 of 31

31 of 31

Acknowledgments

Class Central Classrooms beta

YouTube videos curated by Class Central.

Classroom Contents

Facilitating Electronic Structure Computations on GPU-based Exascale Platforms

Automatically move to the next video in the Classroom when playback concludes

  1. 1 HPC Best Practices Webinar Series
  2. 2 Algorithms and performance portability for electroni structure
  3. 3 Speeding up electronic structure calculations to ena
  4. 4 Running MD on exascale platforms
  5. 5 Main numerical kernels for electronic structure calculations
  6. 6 Developing alternative solvers based on polynomial matrices
  7. 7 Implementation divided into two libraries
  8. 8 Using OpenMP for GPU offloading
  9. 9 General implementation strategy
  10. 10 Computer Science challenges
  11. 11 BML: supported (shared memory) matrix formats
  12. 12 BML: Supporting multiple data types in a C code
  13. 13 BML: Fortran interface is important for targeted application codes
  14. 14 BML: Unit test/Continuous integration
  15. 15 Offloading to GPU
  16. 16 Offloading strategy
  17. 17 GPU offloading with OpenMP
  18. 18 Challenges in interfacing with optimized vendor libra
  19. 19 Using a synthetic Hamiltonian matrix for Performanc Benchmarking
  20. 20 rocSPARSE performance on Crusher @ OLCF
  21. 21 Chebyshev expansions for modest matrix sizes (metals)
  22. 22 Chebyshev expansion of Density Matrix
  23. 23 Exploiting GPU concurrency in calculating Chebysh terms
  24. 24 Distributing computation
  25. 25 Balancing computational cost and accuracy with matrix thresholding
  26. 26 A non-intrusive implementation
  27. 27 What about wavefunction-based solver? (Planewaves...)
  28. 28 Numerical Discretization of DFT problem
  29. 29 Parallel scaling/performance on Summit
  30. 30 Lesson learned: Efficiently using GPUs requires a lo work!
  31. 31 Acknowledgments

Never Stop Learning.

Get personalized course recommendations, track subjects and courses with reminders, and more.

Someone learning on their laptop while sitting on the floor.