portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2024-09-20 01:05:38 +01:00
Code Releases Activity
feature/cpu-threaded-smp
Grid/systems/Tursa/dwf.4node.perf
Peter Boyle 145acf2919 Perf results
2021-09-16 01:06:28 +01:00

246 lines
15 KiB
Plaintext
Raw Permalink Blame History

tu-c0r0n00 - 0 device=0 binding=--interleave=0,1
tu-c0r0n00 - 1 device=1 binding=--interleave=2,3
tu-c0r0n09 - 1 device=1 binding=--interleave=2,3
tu-c0r0n00 - 2 device=2 binding=--interleave=4,5
tu-c0r0n06 - 0 device=0 binding=--interleave=0,1
tu-c0r0n06 - 1 device=1 binding=--interleave=2,3
tu-c0r0n09 - 0 device=0 binding=--interleave=0,1
tu-c0r0n09 - 2 device=2 binding=--interleave=4,5
tu-c0r0n03 - 1 device=1 binding=--interleave=2,3
tu-c0r0n06 - 2 device=2 binding=--interleave=4,5
tu-c0r0n09 - 3 device=3 binding=--interleave=6,7
tu-c0r0n00 - 3 device=3 binding=--interleave=6,7
tu-c0r0n03 - 0 device=0 binding=--interleave=0,1
tu-c0r0n03 - 2 device=2 binding=--interleave=4,5
tu-c0r0n06 - 3 device=3 binding=--interleave=6,7
tu-c0r0n03 - 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[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: ================================================
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 16
SharedMemoryMpi: Node communicator of size 4
0SharedMemoryMpi: SharedMemoryMPI.cc acceleratorAllocDevice 2147483648bytes at 0x7fcd80000000 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 : 1.198523 s : Grid Layout
Grid : Message : 1.198530 s : Global lattice size : 64 64 64 64
Grid : Message : 1.198534 s : OpenMP threads : 4
Grid : Message : 1.198535 s : MPI tasks : 2 2 2 2
Grid : Message : 1.397615 s : Making s innermost grids
Grid : Message : 1.441828 s : Initialising 4d RNG
Grid : Message : 1.547973 s : Intialising parallel RNG with unique string 'The 4D RNG'
Grid : Message : 1.547998 s : Seed SHA256: 49db4542db694e3b1a74bf2592a8c1b83bfebbe18401693c2609a4c3af1
Grid : Message : 1.954777 s : Initialising 5d RNG
Grid : Message : 3.633825 s : Intialising parallel RNG with unique string 'The 5D RNG'
Grid : Message : 3.633869 s : Seed SHA256: b6316f2fac44ce14111f93e0296389330b077bfd0a7b359f781c58589f8a
Grid : Message : 12.162710 s : Initialised RNGs
Grid : Message : 15.882520 s : Drawing gauge field
Grid : Message : 15.816362 s : Random gauge initialised
Grid : Message : 17.279671 s : Setting up Cshift based reference
Grid : Message : 26.331426 s : *****************************************************************
Grid : Message : 26.331452 s : * Kernel options --dslash-generic, --dslash-unroll, --dslash-asm
Grid : Message : 26.331454 s : *****************************************************************
Grid : Message : 26.331456 s : *****************************************************************
Grid : Message : 26.331458 s : * Benchmarking DomainWallFermionR::Dhop
Grid : Message : 26.331459 s : * Vectorising space-time by 8
Grid : Message : 26.331463 s : * VComplexF size is 64 B
Grid : Message : 26.331465 s : * SINGLE precision
Grid : Message : 26.331467 s : * Using Overlapped Comms/Compute
Grid : Message : 26.331468 s : * Using GENERIC Nc WilsonKernels
Grid : Message : 26.331469 s : *****************************************************************
Grid : Message : 28.413717 s : Called warmup
Grid : Message : 56.418423 s : Called Dw 3000 times in 2.80047e+07 us
Grid : Message : 56.418476 s : mflop/s = 3.79581e+07
Grid : Message : 56.418479 s : mflop/s per rank = 2.37238e+06
Grid : Message : 56.418481 s : mflop/s per node = 9.48953e+06
Grid : Message : 56.418483 s : RF GiB/s (base 2) = 77130
Grid : Message : 56.418485 s : mem GiB/s (base 2) = 48206.3
Grid : Message : 56.422076 s : norm diff 1.03481e-13
Grid : Message : 56.456894 s : #### Dhop calls report
Grid : Message : 56.456899 s : WilsonFermion5D Number of DhopEO Calls : 6002
Grid : Message : 56.456903 s : WilsonFermion5D TotalTime /Calls : 4710.93 us
Grid : Message : 56.456905 s : WilsonFermion5D CommTime /Calls : 3196.15 us
Grid : Message : 56.456908 s : WilsonFermion5D FaceTime /Calls : 494.392 us
Grid : Message : 56.456910 s : WilsonFermion5D ComputeTime1/Calls : 44.4107 us
Grid : Message : 56.456912 s : WilsonFermion5D ComputeTime2/Calls : 1037.75 us
Grid : Message : 56.456921 s : Average mflops/s per call : 3.55691e+09
Grid : Message : 56.456925 s : Average mflops/s per call per rank : 2.22307e+08
Grid : Message : 56.456928 s : Average mflops/s per call per node : 8.89228e+08
Grid : Message : 56.456930 s : Average mflops/s per call (full) : 3.82915e+07
Grid : Message : 56.456933 s : Average mflops/s per call per rank (full): 2.39322e+06
Grid : Message : 56.456952 s : Average mflops/s per call per node (full): 9.57287e+06
Grid : Message : 56.456954 s : WilsonFermion5D Stencil
Grid : Message : 56.457016 s : Stencil calls 3001
Grid : Message : 56.457022 s : Stencil halogtime 0
Grid : Message : 56.457024 s : Stencil gathertime 55.9154
Grid : Message : 56.457026 s : Stencil gathermtime 20.1073
Grid : Message : 56.457028 s : Stencil mergetime 18.5585
Grid : Message : 56.457030 s : Stencil decompresstime 0.0639787
Grid : Message : 56.457032 s : Stencil comms_bytes 4.02653e+08
Grid : Message : 56.457034 s : Stencil commtime 6379.93
Grid : Message : 56.457036 s : Stencil 63.1124 GB/s per rank
Grid : Message : 56.457038 s : Stencil 252.45 GB/s per node
Grid : Message : 56.457040 s : WilsonFermion5D StencilEven
Grid : Message : 56.457048 s : WilsonFermion5D StencilOdd
Grid : Message : 56.457062 s : WilsonFermion5D Stencil Reporti()
Grid : Message : 56.457065 s : WilsonFermion5D StencilEven Reporti()
Grid : Message : 56.457066 s : WilsonFermion5D StencilOdd Reporti()
Grid : Message : 79.259261 s : Compare to naive wilson implementation Dag to verify correctness
Grid : Message : 79.259287 s : Called DwDag
Grid : Message : 79.259288 s : norm dag result 12.0421
Grid : Message : 79.271740 s : norm dag ref 12.0421
Grid : Message : 79.287759 s : norm dag diff 7.63236e-14
Grid : Message : 79.328100 s : Calling Deo and Doe and //assert Deo+Doe == Dunprec
Grid : Message : 79.955951 s : src_e0.499997
Grid : Message : 80.633620 s : src_o0.500003
Grid : Message : 80.164163 s : *********************************************************
Grid : Message : 80.164168 s : * Benchmarking DomainWallFermionF::DhopEO
Grid : Message : 80.164170 s : * Vectorising space-time by 8
Grid : Message : 80.164172 s : * SINGLE precision
Grid : Message : 80.164174 s : * Using Overlapped Comms/Compute
Grid : Message : 80.164177 s : * Using GENERIC Nc WilsonKernels
Grid : Message : 80.164178 s : *********************************************************
Grid : Message : 93.797635 s : Deo mflop/s = 3.93231e+07
Grid : Message : 93.797670 s : Deo mflop/s per rank 2.45769e+06
Grid : Message : 93.797672 s : Deo mflop/s per node 9.83077e+06
Grid : Message : 93.797674 s : #### Dhop calls report
Grid : Message : 93.797675 s : WilsonFermion5D Number of DhopEO Calls : 3001
Grid : Message : 93.797677 s : WilsonFermion5D TotalTime /Calls : 4542.83 us
Grid : Message : 93.797679 s : WilsonFermion5D CommTime /Calls : 2978.97 us
Grid : Message : 93.797681 s : WilsonFermion5D FaceTime /Calls : 602.287 us
Grid : Message : 93.797683 s : WilsonFermion5D ComputeTime1/Calls : 67.1416 us
Grid : Message : 93.797685 s : WilsonFermion5D ComputeTime2/Calls : 1004.07 us
Grid : Message : 93.797713 s : Average mflops/s per call : 3.30731e+09
Grid : Message : 93.797717 s : Average mflops/s per call per rank : 2.06707e+08
Grid : Message : 93.797719 s : Average mflops/s per call per node : 8.26827e+08
Grid : Message : 93.797721 s : Average mflops/s per call (full) : 3.97084e+07
Grid : Message : 93.797727 s : Average mflops/s per call per rank (full): 2.48178e+06
Grid : Message : 93.797732 s : Average mflops/s per call per node (full): 9.92711e+06
Grid : Message : 93.797735 s : WilsonFermion5D Stencil
Grid : Message : 93.797746 s : WilsonFermion5D StencilEven
Grid : Message : 93.797758 s : WilsonFermion5D StencilOdd
Grid : Message : 93.797769 s : Stencil calls 3001
Grid : Message : 93.797773 s : Stencil halogtime 0
Grid : Message : 93.797776 s : Stencil gathertime 56.7458
Grid : Message : 93.797780 s : Stencil gathermtime 22.6504
Grid : Message : 93.797782 s : Stencil mergetime 21.1913
Grid : Message : 93.797786 s : Stencil decompresstime 0.0556481
Grid : Message : 93.797788 s : Stencil comms_bytes 2.01327e+08
Grid : Message : 93.797791 s : Stencil commtime 2989.33
Grid : Message : 93.797795 s : Stencil 67.3484 GB/s per rank
Grid : Message : 93.797798 s : Stencil 269.394 GB/s per node
Grid : Message : 93.797801 s : WilsonFermion5D Stencil Reporti()
Grid : Message : 93.797803 s : WilsonFermion5D StencilEven Reporti()
Grid : Message : 93.797805 s : WilsonFermion5D StencilOdd Reporti()
Grid : Message : 93.873429 s : r_e6.02111
Grid : Message : 93.879931 s : r_o6.02102
Grid : Message : 93.885912 s : res12.0421
Grid : Message : 94.876555 s : norm diff 0
Grid : Message : 95.485643 s : norm diff even 0
Grid : Message : 95.581236 s : norm diff odd 0
View Git Blame Copy Permalink