portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2024-11-14 01:35:36 +00:00
Code Releases Activity
b2ccaad761
Grid/systems/Tursa/dwf.16node.perf

294 lines
17 KiB
Plaintext
Raw Normal View History

Perf results
2021-09-16 01:06:28 +01:00
tu-c0r1n00 - 0 device=0 binding=--interleave=0,1
tu-c0r1n00 - 1 device=1 binding=--interleave=2,3
tu-c0r1n00 - 2 device=2 binding=--interleave=4,5
tu-c0r1n00 - 3 device=3 binding=--interleave=6,7
tu-c0r2n21 - 0 device=0 binding=--interleave=0,1
tu-c0r2n06 - 3 device=3 binding=--interleave=6,7
tu-c0r2n12 - 2 device=2 binding=--interleave=4,5
tu-c0r2n21 - 2 device=2 binding=--interleave=4,5
tu-c0r2n21 - 3 device=3 binding=--interleave=6,7
tu-c0r2n21 - 1 device=1 binding=--interleave=2,3
tu-c0r2n06 - 0 device=0 binding=--interleave=0,1
tu-c0r2n06 - 1 device=1 binding=--interleave=2,3
tu-c0r2n12 - 1 device=1 binding=--interleave=2,3
tu-c0r2n06 - 2 device=2 binding=--interleave=4,5
tu-c0r1n12 - 2 device=2 binding=--interleave=4,5
tu-c0r1n12 - 3 device=3 binding=--interleave=6,7
tu-c0r2n12 - 0 device=0 binding=--interleave=0,1
tu-c0r1n12 - 1 device=1 binding=--interleave=2,3
tu-c0r1n12 - 0 device=0 binding=--interleave=0,1
tu-c0r2n12 - 3 device=3 binding=--interleave=6,7
tu-c0r1n18 - 2 device=2 binding=--interleave=4,5
tu-c0r1n18 - 0 device=0 binding=--interleave=0,1
tu-c0r1n18 - 3 device=3 binding=--interleave=6,7
tu-c0r1n18 - 1 device=1 binding=--interleave=2,3
tu-c0r1n06 - 2 device=2 binding=--interleave=4,5
tu-c0r2n09 - 3 device=3 binding=--interleave=6,7
tu-c0r1n06 - 0 device=0 binding=--interleave=0,1
tu-c0r2n15 - 0 device=0 binding=--interleave=0,1
tu-c0r2n09 - 2 device=2 binding=--interleave=4,5
tu-c0r1n06 - 3 device=3 binding=--interleave=6,7
tu-c0r2n15 - 1 device=1 binding=--interleave=2,3
tu-c0r2n15 - 2 device=2 binding=--interleave=4,5
tu-c0r1n06 - 1 device=1 binding=--interleave=2,3
tu-c0r2n15 - 3 device=3 binding=--interleave=6,7
tu-c0r2n09 - 1 device=1 binding=--interleave=2,3
tu-c0r2n09 - 0 device=0 binding=--interleave=0,1
tu-c0r1n09 - 0 device=0 binding=--interleave=0,1
tu-c0r1n09 - 1 device=1 binding=--interleave=2,3
tu-c0r1n09 - 2 device=2 binding=--interleave=4,5
tu-c0r1n09 - 3 device=3 binding=--interleave=6,7
tu-c0r1n21 - 2 device=2 binding=--interleave=4,5
tu-c0r1n21 - 0 device=0 binding=--interleave=0,1
tu-c0r1n21 - 1 device=1 binding=--interleave=2,3
tu-c0r1n15 - 2 device=2 binding=--interleave=4,5
tu-c0r1n21 - 3 device=3 binding=--interleave=6,7
tu-c0r1n15 - 1 device=1 binding=--interleave=2,3
tu-c0r1n15 - 0 device=0 binding=--interleave=0,1
tu-c0r2n03 - 2 device=2 binding=--interleave=4,5
tu-c0r1n15 - 3 device=3 binding=--interleave=6,7
tu-c0r2n03 - 0 device=0 binding=--interleave=0,1
tu-c0r2n00 - 3 device=3 binding=--interleave=6,7
tu-c0r2n03 - 3 device=3 binding=--interleave=6,7
tu-c0r2n00 - 1 device=1 binding=--interleave=2,3
tu-c0r2n00 - 2 device=2 binding=--interleave=4,5
tu-c0r2n18 - 1 device=1 binding=--interleave=2,3
tu-c0r2n00 - 0 device=0 binding=--interleave=0,1
tu-c0r2n18 - 2 device=2 binding=--interleave=4,5
tu-c0r2n18 - 3 device=3 binding=--interleave=6,7
tu-c0r2n18 - 0 device=0 binding=--interleave=0,1
tu-c0r2n03 - 1 device=1 binding=--interleave=2,3
tu-c0r1n03 - 0 device=0 binding=--interleave=0,1
tu-c0r1n03 - 1 device=1 binding=--interleave=2,3
tu-c0r1n03 - 2 device=2 binding=--interleave=4,5
tu-c0r1n03 - 3 device=3 binding=--interleave=6,7
OPENMPI detected
AcceleratorCudaInit: using default device
AcceleratorCudaInit: assume user either uses a) IBM jsrun, or
AcceleratorCudaInit: b) invokes through a wrapping script to set CUDA_VISIBLE_DEVICES, UCX_NET_DEVICES, and numa binding
AcceleratorCudaInit: Configure options --enable-summit, --enable-select-gpu=no
AcceleratorCudaInit: ================================================
OPENMPI detected
AcceleratorCudaInit: using default device
AcceleratorCudaInit: assume user either uses a) IBM jsrun, or
AcceleratorCudaInit: b) invokes through a wrapping script to set CUDA_VISIBLE_DEVICES, UCX_NET_DEVICES, and numa binding
AcceleratorCudaInit: Configure options --enable-summit, --enable-select-gpu=no
AcceleratorCudaInit: ================================================
OPENMPI detected
AcceleratorCudaInit[0]: ========================
AcceleratorCudaInit[0]: Device Number : 0
AcceleratorCudaInit[0]: ========================
AcceleratorCudaInit[0]: Device identifier: NVIDIA A100-SXM4-40GB
AcceleratorCudaInit[0]: totalGlobalMem: 42505273344
AcceleratorCudaInit[0]: managedMemory: 1
AcceleratorCudaInit[0]: isMultiGpuBoard: 0
AcceleratorCudaInit[0]: warpSize: 32
AcceleratorCudaInit[0]: pciBusID: 3
AcceleratorCudaInit[0]: pciDeviceID: 0
AcceleratorCudaInit[0]: maxGridSize (2147483647,65535,65535)
AcceleratorCudaInit: using default device
AcceleratorCudaInit: assume user either uses a) IBM jsrun, or
AcceleratorCudaInit: b) invokes through a wrapping script to set CUDA_VISIBLE_DEVICES, UCX_NET_DEVICES, and numa binding
AcceleratorCudaInit: Configure options --enable-summit, --enable-select-gpu=no
AcceleratorCudaInit: ================================================
OPENMPI detected
AcceleratorCudaInit[0]: ========================
AcceleratorCudaInit[0]: Device Number : 0
AcceleratorCudaInit[0]: ========================
AcceleratorCudaInit[0]: Device identifier: NVIDIA A100-SXM4-40GB
AcceleratorCudaInit[0]: totalGlobalMem: 42505273344
AcceleratorCudaInit[0]: managedMemory: 1
AcceleratorCudaInit[0]: isMultiGpuBoard: 0
AcceleratorCudaInit[0]: warpSize: 32
AcceleratorCudaInit[0]: pciBusID: 3
AcceleratorCudaInit[0]: pciDeviceID: 0
AcceleratorCudaInit[0]: maxGridSize (2147483647,65535,65535)
AcceleratorCudaInit: using default device
AcceleratorCudaInit: assume user either uses a) IBM jsrun, or
AcceleratorCudaInit: b) invokes through a wrapping script to set CUDA_VISIBLE_DEVICES, UCX_NET_DEVICES, and numa binding
AcceleratorCudaInit: Configure options --enable-summit, --enable-select-gpu=no
AcceleratorCudaInit: ================================================
OPENMPI detected
AcceleratorCudaInit: using default device
AcceleratorCudaInit: assume user either uses a) IBM jsrun, or
AcceleratorCudaInit: b) invokes through a wrapping script to set CUDA_VISIBLE_DEVICES, UCX_NET_DEVICES, and numa binding
AcceleratorCudaInit: Configure options --enable-summit, --enable-select-gpu=no
AcceleratorCudaInit: ================================================
OPENMPI detected
AcceleratorCudaInit: using default device
AcceleratorCudaInit: assume user either uses a) IBM jsrun, or
AcceleratorCudaInit: b) invokes through a wrapping script to set CUDA_VISIBLE_DEVICES, UCX_NET_DEVICES, and numa binding
AcceleratorCudaInit: Configure options --enable-summit, --enable-select-gpu=no
AcceleratorCudaInit: ================================================
OPENMPI detected
AcceleratorCudaInit: using default device
AcceleratorCudaInit: assume user either uses a) IBM jsrun, or
AcceleratorCudaInit: b) invokes through a wrapping script to set CUDA_VISIBLE_DEVICES, UCX_NET_DEVICES, and numa binding
AcceleratorCudaInit: Configure options --enable-summit, --enable-select-gpu=no
AcceleratorCudaInit: ================================================
OPENMPI detected
AcceleratorCudaInit: using default device
AcceleratorCudaInit: assume user either uses a) IBM jsrun, or
AcceleratorCudaInit: b) invokes through a wrapping script to set CUDA_VISIBLE_DEVICES, UCX_NET_DEVICES, and numa binding
AcceleratorCudaInit: Configure options --enable-summit, --enable-select-gpu=no
AcceleratorCudaInit: ================================================
SharedMemoryMpi: World communicator of size 64
SharedMemoryMpi: Node communicator of size 4
0SharedMemoryMpi: SharedMemoryMPI.cc acceleratorAllocDevice 2147483648bytes at 0x7f05c0000000 for comms buffers
Setting up IPC
__|__|__|__|__|__|__|__|__|__|__|__|__|__|__
__|__|__|__|__|__|__|__|__|__|__|__|__|__|__
__|_ | | | | | | | | | | | | _|__
__|_ _|__
__|_ GGGG RRRR III DDDD _|__
__|_ G R R I D D _|__
__|_ G R R I D D _|__
__|_ G GG RRRR I D D _|__
__|_ G G R R I D D _|__
__|_ GGGG R R III DDDD _|__
__|_ _|__
__|__|__|__|__|__|__|__|__|__|__|__|__|__|__
__|__|__|__|__|__|__|__|__|__|__|__|__|__|__
| | | | | | | | | | | | | |
Copyright (C) 2015 Peter Boyle, Azusa Yamaguchi, Guido Cossu, Antonin Portelli and other authors
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
Current Grid git commit hash=9d2238148c56e3fbadfa95dcabf2b83d4bde14cd: (HEAD -> develop) uncommited changes
Grid : Message : ================================================
Grid : Message : MPI is initialised and logging filters activated
Grid : Message : ================================================
Grid : Message : Requested 2147483648 byte stencil comms buffers
Grid : Message : MemoryManager Cache 34004218675 bytes
Grid : Message : MemoryManager::Init() setting up
Grid : Message : MemoryManager::Init() cache pool for recent allocations: SMALL 32 LARGE 8
Grid : Message : MemoryManager::Init() Non unified: Caching accelerator data in dedicated memory
Grid : Message : MemoryManager::Init() Using cudaMalloc
Grid : Message : 1.814936 s : Grid Layout
Grid : Message : 1.814947 s : Global lattice size : 64 64 64 256
Grid : Message : 1.814952 s : OpenMP threads : 4
Grid : Message : 1.814955 s : MPI tasks : 2 2 2 8
Grid : Message : 1.859229 s : Making s innermost grids
Grid : Message : 1.907983 s : Initialising 4d RNG
Grid : Message : 1.999619 s : Intialising parallel RNG with unique string 'The 4D RNG'
Grid : Message : 1.999657 s : Seed SHA256: 49db4542db694e3b1a74bf2592a8c1b83bfebbe18401693c2609a4c3af1
Grid : Message : 3.786102 s : Initialising 5d RNG
Grid : Message : 5.361999 s : Intialising parallel RNG with unique string 'The 5D RNG'
Grid : Message : 5.362036 s : Seed SHA256: b6316f2fac44ce14111f93e0296389330b077bfd0a7b359f781c58589f8a
Grid : Message : 38.698345 s : Initialised RNGs
Grid : Message : 42.821728 s : Drawing gauge field
Grid : Message : 43.916364 s : Random gauge initialised
Grid : Message : 46.410003 s : Setting up Cshift based reference
Grid : Message : 54.242661 s : *****************************************************************
Grid : Message : 54.242686 s : * Kernel options --dslash-generic, --dslash-unroll, --dslash-asm
Grid : Message : 54.242688 s : *****************************************************************
Grid : Message : 54.242689 s : *****************************************************************
Grid : Message : 54.242690 s : * Benchmarking DomainWallFermionR::Dhop
Grid : Message : 54.242691 s : * Vectorising space-time by 8
Grid : Message : 54.242692 s : * VComplexF size is 64 B
Grid : Message : 54.242694 s : * SINGLE precision
Grid : Message : 54.242697 s : * Using Overlapped Comms/Compute
Grid : Message : 54.242698 s : * Using GENERIC Nc WilsonKernels
Grid : Message : 54.242699 s : *****************************************************************
Grid : Message : 56.314112 s : Called warmup
Grid : Message : 84.246354 s : Called Dw 3000 times in 2.79318e+07 us
Grid : Message : 84.246405 s : mflop/s = 1.52229e+08
Grid : Message : 84.246408 s : mflop/s per rank = 2.37857e+06
Grid : Message : 84.246412 s : mflop/s per node = 9.51428e+06
Grid : Message : 84.246414 s : RF GiB/s (base 2) = 309325
Grid : Message : 84.246417 s : mem GiB/s (base 2) = 193328
Grid : Message : 84.250016 s : norm diff 1.03478e-13
Grid : Message : 84.285132 s : #### Dhop calls report
Grid : Message : 84.285137 s : WilsonFermion5D Number of DhopEO Calls : 6002
Grid : Message : 84.285140 s : WilsonFermion5D TotalTime /Calls : 4703.27 us
Grid : Message : 84.285142 s : WilsonFermion5D CommTime /Calls : 3131.05 us
Grid : Message : 84.285144 s : WilsonFermion5D FaceTime /Calls : 492.972 us
Grid : Message : 84.285146 s : WilsonFermion5D ComputeTime1/Calls : 56.9085 us
Grid : Message : 84.285148 s : WilsonFermion5D ComputeTime2/Calls : 1099.95 us
Grid : Message : 84.285160 s : Average mflops/s per call : 1.43412e+10
Grid : Message : 84.285165 s : Average mflops/s per call per rank : 2.24082e+08
Grid : Message : 84.285170 s : Average mflops/s per call per node : 8.96328e+08
Grid : Message : 84.285173 s : Average mflops/s per call (full) : 1.53416e+08
Grid : Message : 84.285176 s : Average mflops/s per call per rank (full): 2.39712e+06
Grid : Message : 84.285194 s : Average mflops/s per call per node (full): 9.58847e+06
Grid : Message : 84.285197 s : WilsonFermion5D Stencil
Grid : Message : 84.285271 s : Stencil calls 3001
Grid : Message : 84.285275 s : Stencil halogtime 0
Grid : Message : 84.285277 s : Stencil gathertime 55.2059
Grid : Message : 84.285281 s : Stencil gathermtime 20.0923
Grid : Message : 84.285283 s : Stencil mergetime 18.9057
Grid : Message : 84.285286 s : Stencil decompresstime 0.0619793
Grid : Message : 84.285289 s : Stencil comms_bytes 4.02653e+08
Grid : Message : 84.285292 s : Stencil commtime 6323.57
Grid : Message : 84.285295 s : Stencil 63.675 GB/s per rank
Grid : Message : 84.285298 s : Stencil 254.7 GB/s per node
Grid : Message : 84.285301 s : WilsonFermion5D StencilEven
Grid : Message : 84.285316 s : WilsonFermion5D StencilOdd
Grid : Message : 84.285333 s : WilsonFermion5D Stencil Reporti()
Grid : Message : 84.285336 s : WilsonFermion5D StencilEven Reporti()
Grid : Message : 84.285337 s : WilsonFermion5D StencilOdd Reporti()
Grid : Message : 106.985790 s : Compare to naive wilson implementation Dag to verify correctness
Grid : Message : 106.985814 s : Called DwDag
Grid : Message : 106.985815 s : norm dag result 12.0421
Grid : Message : 107.188790 s : norm dag ref 12.0421
Grid : Message : 107.349010 s : norm dag diff 7.63254e-14
Grid : Message : 107.762980 s : Calling Deo and Doe and //assert Deo+Doe == Dunprec
Grid : Message : 107.458374 s : src_e0.499998
Grid : Message : 107.754073 s : src_o0.500002
Grid : Message : 107.855191 s : *********************************************************
Grid : Message : 107.855194 s : * Benchmarking DomainWallFermionF::DhopEO
Grid : Message : 107.855195 s : * Vectorising space-time by 8
Grid : Message : 107.855197 s : * SINGLE precision
Grid : Message : 107.855198 s : * Using Overlapped Comms/Compute
Grid : Message : 107.855199 s : * Using GENERIC Nc WilsonKernels
Grid : Message : 107.855200 s : *********************************************************
Grid : Message : 121.549348 s : Deo mflop/s = 1.56492e+08
Grid : Message : 121.549382 s : Deo mflop/s per rank 2.44518e+06
Grid : Message : 121.549384 s : Deo mflop/s per node 9.78074e+06
Grid : Message : 121.549387 s : #### Dhop calls report
Grid : Message : 121.549388 s : WilsonFermion5D Number of DhopEO Calls : 3001
Grid : Message : 121.549390 s : WilsonFermion5D TotalTime /Calls : 4563.01 us
Grid : Message : 121.549393 s : WilsonFermion5D CommTime /Calls : 2967.77 us
Grid : Message : 121.549395 s : WilsonFermion5D FaceTime /Calls : 601.095 us
Grid : Message : 121.549397 s : WilsonFermion5D ComputeTime1/Calls : 59.9877 us
Grid : Message : 121.549399 s : WilsonFermion5D ComputeTime2/Calls : 1038.46 us
Grid : Message : 121.549423 s : Average mflops/s per call : 1.2726e+10
Grid : Message : 121.549428 s : Average mflops/s per call per rank : 1.98843e+08
Grid : Message : 121.549430 s : Average mflops/s per call per node : 7.95373e+08
Grid : Message : 121.549432 s : Average mflops/s per call (full) : 1.58131e+08
Grid : Message : 121.549436 s : Average mflops/s per call per rank (full): 2.4708e+06
Grid : Message : 121.549440 s : Average mflops/s per call per node (full): 9.88321e+06
Grid : Message : 121.549442 s : WilsonFermion5D Stencil
Grid : Message : 121.549453 s : WilsonFermion5D StencilEven
Grid : Message : 121.549472 s : WilsonFermion5D StencilOdd
Grid : Message : 121.549484 s : Stencil calls 3001
Grid : Message : 121.549490 s : Stencil halogtime 0
Grid : Message : 121.549492 s : Stencil gathertime 55.2206
Grid : Message : 121.549496 s : Stencil gathermtime 19.4562
Grid : Message : 121.549500 s : Stencil mergetime 18.3469
Grid : Message : 121.549502 s : Stencil decompresstime 0.0646451
Grid : Message : 121.549506 s : Stencil comms_bytes 2.01327e+08
Grid : Message : 121.549510 s : Stencil commtime 2979.17
Grid : Message : 121.549512 s : Stencil 67.5782 GB/s per rank
Grid : Message : 121.549514 s : Stencil 270.313 GB/s per node
Grid : Message : 121.549517 s : WilsonFermion5D Stencil Reporti()
Grid : Message : 121.549519 s : WilsonFermion5D StencilEven Reporti()
Grid : Message : 121.549522 s : WilsonFermion5D StencilOdd Reporti()
Grid : Message : 121.625928 s : r_e6.02108
Grid : Message : 121.634489 s : r_o6.02101
Grid : Message : 121.640496 s : res12.0421
Grid : Message : 122.275455 s : norm diff 0
Grid : Message : 123.135840 s : norm diff even 0
Grid : Message : 123.389190 s : norm diff odd 0
Copy Permalink