CUDA Libraries and MPI+OpenMP+CUDA - Prace Training Portal

More documents

Recommendations

Info

Time-consuming steps in PWSCF• Calculation of Charge Density– FFT + matrix-matrix multiplication• Calculation of Potential– FFT + operations on real-space grid• Davidson Iterative Diagonalization (SCF)– FFT + eigenvalues/eigenvectors problem + matrix-matrix multiplicationBasically most CPU time spent in linear-algebra operations,implemented in BLAS and LAPACK libraries, and in FFT!February 10, 2012 PRACE Winter School 201240
Parallelization levels in PWSCFImages• Only for Nudged Elastic Band (NEB) calculationsGPU!K-pointsPlane-wavesLinear algebra &task groupsMulti-threadedkernels• Distribution over k-points (if more than one)• Scalability over k-points (if more than one)• No memory scaling• Distribution of wave-function coefficients• Distribution of real-grid points• Good memory scale, good overall scalability, LB• Iterative diagonalization (fully-parallel or serial)• Smart grouping of 3DFFTs to reduce compulsoryMPI communications• OpenMP handled explicitly or implicitly• Extend the scaling on multi-core machines with“limited” memoryFebruary 10, 2012 PRACE Winter School 201241
Page 1 and 2: PRACE Winter School 2012Hybrid Prog
Page 3 and 4: Acknowledgments first!• ICHEC, Ir
Page 5 and 6: CUDA library ecosystemNVIDIA CUDATo
Page 7 and 8: CUFFT - Code Sample (C)#define NX 2
Page 9 and 10: CUBLAS - Thunking versus non-Thunki
Page 11 and 12: CUBLAS - Code Sample (FORTRAN)PROGR
Page 13 and 14: use-case: compute 3D-FFT on multi-G
Page 15 and 16: fft3Dtest.cu - Code Snippet (C++)lo
Page 17 and 18: fft3Dtest.cu - Profiling12101 GPU2
Page 19 and 20: Pinned allocation - Code Sample (FO
Page 21 and 22: Pinned allocation - Code Sample (FO
Page 23 and 24: Consideration over PINNED memory4m3
Page 25: PHIGEMM - Code Snippet (C)for (iDev
Page 28 and 29: cuBLAS API: new versus Legacy• th
Page 30 and 31: Performance Measurement and Metrics
Page 32 and 33: Performance Measurement and Metrics
Page 34 and 35: MPI + CUDA, short summary1MPI commu
Page 36 and 37: Know what you have…Let’s forget
Page 38 and 39: Example: PLASMA, DPLASMA & MAGMAMAG
Page 42 and 43: GPU Developments• MPI-GPU binding
Page 44 and 45: MPI-GPU binding - Code Snippet (C)/
Page 46 and 47: GPU memory management - Code Snippe
Page 48 and 49: ADDUSDENS - Code Snippet (FORTRAN)D
Page 50 and 51: QVAN2 - Code Snippet (FORTRAN)DO lm
Page 52 and 53: CUDA ADDUSDENS, Considerations• C
Page 54 and 55: “new” CUDA ADDUSDENS - Code Sni
Page 56 and 57: CUDA NEWD - Code Snippet (C)for( ih
Page 58 and 59: CUDA NEWD - Code Snippet (C)...#if
Page 60 and 61: CUDA VLOC_PSI_K serial - Computatio
Page 62 and 63: KERNEL_INIT_PSIC_K - Code Snippet (
Page 64 and 65: CUDA VLOC_PSI_K - parallelPSIPSICFF
Page 66 and 67: PWSCF GPU, results & benchmarks(ser
Page 68 and 69: WHEN and WHY more MPI processesper
Page 70: Thank you for your attention!No CPU

CUDA Libraries and MPI+OpenMP+CUDA - Prace Training Portal

Create successful ePaper yourself

Delete template?

Save as template?