# Invocation command line: # /data/caar/spec/hpc2021-1.0.2/bin/harness/runhpc --configfile gcc11_romeSocketPlaces.cfg --reportable --tune base --ranks 2 --threads 64 --nopower --runmode speed --tune base --size ref tiny # output_root was not used for this run ############################################################################ ###################################################################### # Example configuration file for the GNU Compilers # # Defines: "model" => "mpi", "acc", "omp", "tgt", "tgtgpu" default "mpi" # "label" => ext base label, default "nv" # # MPI-only Command: # runhpc -c Example_gnu --reportable -T base --define model=mpi --ranks=40 small # # OpenACC Command: # runhpc -c Example_gnu --reportable -T base --define model=acc --ranks=4 small # # OpenMP Command: # runhpc -c Example_gnu --reportable -T base --define model=omp --ranks=1 --threads=40 small # # OpenMP Target Offload to Host Command: # runhpc -c Example_gnu --reportable -T base --define model=tgt --ranks=1 --threads=40 small # # OpenMP Target Offload to NVIDIA GPU Command: # runhpc -c Example_gnu --reportable -T base --define model=tgtnv --ranks=4 small # ####################################################################### %ifndef %{label} # IF label is not set use gnu % define label gcc11_romeSocketPlaces %endif %ifndef %{model} # IF model is not set use mpi % define model omp %endif teeout = yes makeflags=-j 80 # Tester Information license_num = 065A test_sponsor = Helmholtz-Zentrum Dresden - Rossendorf tester = Helmholtz-Zentrum Dresden - Rossendorf ###################################################### # SUT Section ###################################################### #include: Example_SUT.inc # ----- Begin inclusion of 'Example_SUT.inc' ############################################################################ ###################################################### # Example configuration information for a # system under test (SUT) Section ###################################################### # General SUT info system_vendor = NEC system_name = Hemera: GIGABYTE H262-Z61 (AMD EPYC 7702) hw_avail = Aug-2019 sw_avail = Jul-2021 prepared_by = Jeffrey Kelling # Computation node info # [Node_Description: Hardware] node_compute_syslbl = Compute Node node_compute_order = 1 node_compute_count = 1 node_compute_purpose = compute node_compute_hw_vendor = Gigabyte node_compute_hw_model = H262-Z61 node_compute_hw_cpu_name = AMD EPYC 7702 node_compute_hw_ncpuorder = 1 or 2 chips per node node_compute_hw_nchips = 2 node_compute_hw_ncores = 64 node_compute_hw_ncoresperchip = 64 node_compute_hw_nthreadspercore = 1 node_compute_hw_cpu_char = Max Boost Clock up to 3.35 GHz node_compute_hw_cpu_mhz = 2000 node_compute_hw_pcache = 32 KB I + 32 KB D on chip per core node_compute_hw_scache = 512 KB I+D on chip per core node_compute_hw_tcache000= 256 MB I+D on chip per chip node_compute_hw_tcache001 = 16 MB shared / 4 cores node_compute_hw_ocache = None node_compute_hw_memory = 512 GB (16 x 32GB 2Rx4 PC4-3200AA-RB2-12-RB0) node_compute_hw_disk = 1 x 500 GB SSD node_compute_hw_other = None #[Node_Description: Accelerator] node_compute_hw_accel_model = -- node_compute_hw_accel_count = 0 node_compute_hw_accel_vendor= -- node_compute_hw_accel_type = -- node_compute_hw_accel_connect = -- node_compute_hw_accel_ecc = -- node_compute_hw_accel_desc = -- #[Node_Description: Software] node_compute_hw_adapter_fs_model = Mellanox MT4119 node_compute_hw_adapter_fs_count = 2 node_compute_hw_adapter_fs_slot_type = PCIe 4.0 16x node_compute_hw_adapter_fs_data_rate = 100 Gb/s node_compute_hw_adapter_fs_ports_used = 2 node_compute_hw_adapter_fs_interconnect = EDR Infiniband node_compute_hw_adapter_fs_driver = -- node_compute_hw_adapter_fs_firmware = 16.26.1040 node_compute_sw_os000 = CentOS Linux release 7.9.2009 (Core) node_compute_sw_os001 = 3.10.0-1160.6.1.el7.x86_64 node_compute_sw_localfile = xfs node_compute_sw_sharedfile000= GPFS Version 5.0.5.0 node_compute_sw_sharedfile001 = 6 NSD (vendor: NEC) node_compute_sw_sharedfile002 = 5 building blocks (vendor: NetApp): node_compute_sw_sharedfile003 = 2x (240 x 8 TB HDD) node_compute_sw_sharedfile004 = 1x (180 x 12 TB HDD) node_compute_sw_sharedfile005 = 1x (240 x 16 TB HDD) node_compute_sw_sharedfile006 = 1x (120 x 16 TB HDD) node_compute_sw_state = Multi-user, run level 3 node_compute_sw_other = None #[Fileserver] #[Interconnect] interconnect_fs_syslbl = Infiniband (EDR) interconnect_fs_order = 1 interconnect_fs_purpose = MPI Traffic, GPFS interconnect_fs_hw_vendor = Mellanox Technologies interconnect_fs_hw_model = Mellanox SB7790 interconnect_fs_hw_switch_fs_model000= 36 x EDR 100 Gb/s interconnect_fs_hw_switch_fs_count = 2 interconnect_fs_hw_switch_fs_ports = 36 interconnect_fs_hw_topo = Mesh (blocking factor: 8:1) interconnect_fs_hw_switch_fs_data_rate = 100 Gb/s interconnect_fs_hw_switch_fs_firmware = -- ####################################################################### # End of SUT section # If this config file were to be applied to several SUTs, edits would # be needed only ABOVE this point. ###################################################################### # ---- End inclusion of '/data/caar/spec/hpc2021-1.0.2/config/Example_SUT.inc' #[Software] sw_compiler000 = C/C++/Fortran: Version 11.2 of sw_compiler001 = GNU Compilers sw_mpi_library = OpenMPI Version 4.0.4 sw_mpi_other = None system_class = Homogenous Cluster sw_other = None #[General notes] notes_005 =OMP_PLACES = "{0:4}:32:4" notes_010 =OMP_PROC_BIND = "true" ####################################################################### # End of SUT section ###################################################################### ###################################################################### # The header section of the config file. Must appear # before any instances of "section markers" (see below) # # ext = how the binaries you generated will be identified # tune = specify "base" or "peak" or "all" label = %{label}_%{model} tune = base output_format = text use_submit_for_speed = 1 # Compiler Settings default: CC = mpicc CXX = mpicxx FC = mpif90 # Compiler Version Flags CC_VERSION_OPTION = --version CXX_VERSION_OPTION = --version FC_VERSION_OPTION = --version %if %{model} eq 'omp' preENV_OMP_PROC_BIND=true preENV_OMP_PLACES={0:4}:32:4 %endif # MPI options and binding environment, dependent upon Model being run # Adjust to match your system MPIRUN_OPTS = --bind-to socket submit = mpirun ${MPIRUN_OPTS} -np $ranks $command flagsurl000=http://www.spec.org/hpc2021/flags/gcc.2021-10-28.xml ####################################################################### # Optimization # Note that SPEC baseline rules require that all uses of a given compiler # use the same flags in the same order. See the SPEChpc Run Rules # for more details # http://www.spec.org/hpc2021/Docs/runrules.html # # OPTIMIZE = flags applicable to all compilers # FOPTIMIZE = flags appliable to the Fortran compiler # COPTIMIZE = flags appliable to the C compiler # CXXOPTIMIZE = flags appliable to the C++ compiler # # See your compiler manual for information on the flags available # for your compiler # Compiler flags applied to all models default=base=default: COPTIMIZE = -Ofast -march=native -lm # use -mcpu=native for ARM CXXOPTIMIZE = -Ofast -march=native -std=c++14 FOPTIMIZE = -Ofast -march=native -ffree-line-length-none -fno-stack-protector #FPPPORTABILITY += -DSPEC_USE_MPIFH -I${MPI_ROOT}/include/ %if %{model} eq 'mpi' pmodel=MPI %endif # OpenACC flags %if %{model} eq 'acc' pmodel=ACC OPTIMIZE += -fopenacc -foffload=-lm %endif # OpenMP (CPU) flags %if %{model} eq 'omp' pmodel=OMP OPTIMIZE += -fopenmp %endif # OpenMP Targeting host flags %if %{model} eq 'tgt' pmodel=TGT OPTIMIZE += -fopenmp %endif # OpenMP Targeting Nvidia GPU flags %if %{model} eq 'tgtnv' pmodel=TGT OPTIMIZE += -fopenmp -fopenmp-targets=nvptx64-nvidia-cuda %endif # No peak flags set, so make peak use the same flags as base default=peak=default: basepeak=1 ####################################################################### # Portability ####################################################################### # The following section was added automatically, and contains settings that # did not appear in the original configuration file, but were added to the # raw file after the run. default: notes_000 =Environment variables set by runhpc before the start of the run: notes_submit_000 =The config file option 'submit' was used. notes_submit_005 = MPI startup command: notes_submit_010 = mpirun --bind-to socket -np $ranks $command