1. Parallel Molecular Dynamics A case study : Programming for performance Laxmikant Kale http://charm.cs.uiuc.edu

2. Molecular Dynamics Collection of [charged] atoms, with bonds Newtonian mechanics At each time-step –Calculate forces on each atom bonds: non-bonded: electrostatic and van der Waal’s –Calculate velocities and Advance positions 1 femtosecond time-step, millions needed! Thousands of atoms (1,000 - 100,000)

3. Further MD Use of cut-off radius to reduce work –8 - 14 Å –Faraway charges ignored! 80-95 % work is non-bonded force computations Some simulations need faraway contributions

4. Scalability The Program should scale up to use a large number of processors. –But what does that mean? An individual simulation isn’t truly scalable Better definition of scalability: –If I double the number of processors, I should be able to retain parallel efficiency by increasing the problem size

5. Isoefficiency Quantify scalability How much increase in problem size is needed to retain the same efficiency on a larger machine? Efficiency : Seq. Time/ (P · Parallel Time) –parallel time = computation + communication + idle

6. Traditional Approaches Replicated Data: –All atom coordinates stored on each processor –Non-bonded Forces distributed evenly –Analysis: Assume N atoms, P processors Computation: O(N/P) Communication: O(N log P) Communication/Computation ratio: P log P Fraction of communication increases with number of processors, independent of problem size!

7. Atom decomposition Partition the Atoms array across processors –Nearby atoms may not be on the same processor –Communication: O(N) per processor –Communication/Computation: O(P)

8. Force Decomposition Distribute force matrix to processors –Matrix is sparse, non uniform –Each processor has one block –Communication: N/sqrt(P) –Ratio: sqrt(P) Better scalability (can use 100+ processors) –Hwang, Saltz, et al: –6% on 32 Pes 36% on 128 processor

9. Spatial Decomposition Allocate close-by atoms to the same processor Three variations possible: –Partitioning into P boxes, 1 per processor Good scalability, but hard to implement –Partitioning into fixed size boxes, each a little larger than the cutoff disctance –Partitioning into smaller boxes Communication: O(N/P)

10. Spatial Decomposition in NAMD NAMD 1 used spatial decomposition Good theoretical isoefficiency, but for a fixed size system, load balancing problems For midsize systems, got good speedups up to 16 processors…. Use the symmetry of Newton’s 3rd law to facilitate load balancing

11. Spatial Decomposition

13. FD + SD Now, we have many more objects to load balance: –Each diamond can be assigned to any processor – Number of diamonds (3D): –14·Number of Patches

14. Bond Forces Multiple types of forces: –Bonds(2), Angles(3), Dihedrals (4),.. –Luckily, each involves atoms in neighboring patches only Straightforward implementation: –Send message to all neighbors, –receive forces from them –26*2 messages per patch!

15. Bonded Forces: Assume one patch per processor B CA

16. Implementation Multiple Objects per processor –Different types: patches, pairwise forces, bonded forces, –Each may have its data ready at different times –Need ability to map and remap them –Need prioritized scheduling Charm++ supports all of these

17. Charm++ Data Driven Objects Object Groups: –global object with a “representative” on each PE Asynchronous method invocation Prioritized scheduling Mature, robust, portable http://charm.cs.uiuc.edu

18. Data driven execution Scheduler Message Q

19. Load Balancing Is a major challenge for this application –especially for a large number of processors Unpredictable workloads –Each diamond (force object) and patch encapsulate variable amount of work –Static estimates are inaccurate Measurement based Load Balancing –Very slow variations across timesteps

20. Bipartite graph balancing Background load: –Patches and angle forces Migratable load: –Non-bonded forces Bipartite communication graph –between migratable and non-migratable objects Challenge: –Balance Load while minimizing communication

21. Load balancing Collect timing data for several cycles Run heuristic load balancer –Several alternative ones Re-map and migrate objects accordingly –Registration mechanisms facilitate migration

23. Performance: size of system

24. Performance: various machines

25. Speedup