# Invocation command line: # /autobench/sources/speccpu2006/bin/runspec --config speccpu.cfg --action run --rate 1 --copies 128 --reportable --nobuild --tune all --iterations 3 --define 32c750=1 --define benchmark=fp --define sles11sp2=1 --define mybasearchp7=1 --define mybaseopt=5 --define libhuge=1 --define smartheap=1 --extension xlc12-sles11sp2-121128 fp # output_root was not used for this run ############################################################################ ####################################################################### # # SPEC CPU2006 Benchmark Config file IBM eServer pSeries Linux # ####################################################################### # THIS CONFIG FILE IS FOR BUILDING on P7+ # # --define (-S) options that are passed to the runspec command # that dynamically affect this config file # # numcores=n The number of cores in the system # to determine how many copies to run in rate # # mybaseopt=x Allows override where x=3,4,5 (-O3, -O4, -O5) for base runs # # nolibhuge Allows override to not use extra libraries on base runs # nosmartheap # # debug Adds debug ('-g') compiler option # mymode64 ##################################################################### # System Under Test (SUT) Section # Replace the setting below with the ones that match your # machine. ##################################################################### sysinfo_program = # System Description %ifdef %{48c760} %ifdef %{rhel64} hw_model = IBM Power 760 (3.4 GHz, 48 core, RHEL) %endif %ifdef %{sles11sp2} hw_model = IBM Power 760 (3.4 GHz, 48 core, SLES) %endif hw_avail = Mar-2013 hw_memory = 512 GB (64x8 GB) DDR3 1066 MHz hw_disk = 3 x 146.8 GB Raid0 SAS SFF 15K RPM hw_cpu_mhz = 3416 hw_nchips = 4 hw_ncores = 48 hw_ncoresperchip = 12 hw_nthreadspercore = 4 hw_fpu = Integrated hw_ncpuorder = 12, 24, 36, 48 cores hw_ocache = None hw_other = None hw_cpu_char000 = Intelligent Energy Optimization enabled, hw_cpu_char001 = up to 3.787 GHz %endif %ifdef %{32c750} %ifdef %{rhel64} hw_model = IBM Power 750 Express (4.0 GHz, 32 core, RHEL) %endif %ifdef %{sles11sp2} hw_model = IBM Power 750 Express (4.0 GHz, 32 core, SLES) %endif hw_avail = Mar-2013 hw_memory = 256 GB (64 x 4 GB) DDR3 1066 MHz hw_disk = 3 x 146.8 GB Raid0 SAS SFF 15K RPM hw_cpu_mhz = 4060 hw_nchips = 8 hw_ncores = 32 hw_ncoresperchip = 4 hw_nthreadspercore = 4 hw_fpu = Integrated hw_ncpuorder = 8, 16, 24, 32 cores hw_ocache = None hw_other = None hw_cpu_char000 = Intelligent Energy Optimization enabled, hw_cpu_char001 = up to 4.431 GHz %endif %ifdef %{16c730} %ifdef %{rhel64} hw_model = IBM Power 730 Express (4.2 GHz, 16 core, RHEL) %endif %ifdef %{sles11sp2} hw_model = IBM Power 730 Express (4.2 GHz, 16 core, SLES) %endif hw_avail = Feb-2013 hw_memory = 128 GB (16x8 GB) DDR3 1066 MHz hw_disk = 5 x 146.8 GB Raid0 SAS SFF 15K RPM hw_cpu_mhz = 4228 hw_nchips = 2 hw_ncores = 16 hw_ncoresperchip = 8 hw_nthreadspercore = 4 hw_fpu = Integrated hw_ncpuorder = 8, 12, 16 cores hw_ocache = None hw_other = None hw_cpu_char000 = Intelligent Energy Optimization enabled, hw_cpu_char001 = up to 4.540 GHz %endif %ifdef %{16c740} %ifdef %{rhel64} hw_model = IBM Power 740 (4.2 GHz, 16 core, RHEL) %endif %ifdef %{sles11sp2} hw_model = IBM Power 740 (4.2 GHz, 16 core, SLES) %endif hw_avail = Feb-2013 hw_memory = 128 GB (32x4 GB) DDR3 1066 MHz hw_disk = 5 x 146.8 GB Raid0 SAS SFF 15K RPM hw_cpu_mhz = 4228 hw_nchips = 2 hw_ncores = 16 hw_ncoresperchip = 8 hw_nthreadspercore = 4 hw_fpu = Integrated hw_ncpuorder = 6, 8, 12, 16 cores hw_ocache = None hw_other = None hw_cpu_char000 = Intelligent Energy Optimization enabled, hw_cpu_char001 = up to 4.540 GHz %endif %ifdef %{16c7R2} %ifdef %{rhel64} hw_model = IBM PowerLinux 7R2 (4.2 GHz, 16 core, RHEL) %endif %ifdef %{sles11sp2} hw_model = IBM PowerLinux 7R2 (4.2 GHz, 16 core, SLES) %endif hw_avail = Feb-2013 hw_memory = 128 GB (16x8 GB) DDR3 1066 MHz hw_disk = 5 x 146.8 GB Raid0 SAS SFF 15K RPM hw_cpu_mhz = 4228 hw_nchips = 2 hw_ncores = 16 hw_ncoresperchip = 8 hw_nthreadspercore = 4 hw_fpu = Integrated hw_ncpuorder = 16 cores hw_ocache = None hw_other = None hw_cpu_char000 = Intelligent Energy Optimization enabled, hw_cpu_char001 = up to 4.540 GHz %endif hw_vendor = IBM Corporation # CPU description hw_cpu_name = POWER7+ hw_pcache = 32 KB I + 32 KB D on chip per core hw_scache = 256 KB I+D on chip per core hw_tcache = 10 MB I+D on chip per core # Tester description - edit the date, put your license number here! license_num = 11 prepared_by = -- test_sponsor = IBM Corporation tester = IBM Corporation # Operating system, file system sw_avail = Dec-2012 %ifdef %{rhel64} sw_os000 = Red Hat Enterprise Linux Server release 6.4 sw_os001 = (ppc64) kernel sw_file = ext4 %endif %ifdef %{sles11sp2} sw_os000 = SUSE Linux Enterprise Server 11 SP2 sw_os001 = (ppc64) kernel 3.0.42-0.7-ppc64 sw_file = ext3 %endif sw_compiler000 = C/C++: Version 12.1 of IBM XL C/C++ for Linux; %if '%{benchmark}' eq 'fp' sw_compiler001 = Fortran: Version 14.1 of IBM XL Fortran for Linux %endif notes_comp_iler_000 = C/C++ compiler updated to December 2012 PTF notes_comp_iler_005 = Version: 12.01.0000.0002 %if '%{benchmark}' eq 'fp' notes_comp_iler_010 = Fortran compiler updated to December 2012 PTF notes_comp_iler_015 = Version: 14.01.0000.0002 %endif sw_other000 = -Post-Link Optimization for Linux on sw_other001 = POWER, version 5.6.1-7 sw_other002 = -MicroQuill SmartHeap 9 %if '%{benchmark}' eq 'fp' sw_other003 = -Apache C++ Standard Library V4.2.1 %endif sw_base_ptrsize = 32-bit sw_peak_ptrsize = 32/64-bit sw_parallel_other = -- sw_state = Run level 3 (multi-user) notes_submit_000 = The config file option 'submit' was used notes_submit_005 = to assign benchmark copy to specific kernel thread using notes_submit_010 = the "numactl" command (see flags file for details). notes_os_000 = ulimit -s (stack) set to 1048576. notes_os_005 = notes_os_010 = Large pages reserved as follows by root user: %if defined(%{16c730}) || defined(%{16c740}) || defined(%{16c7R2}) notes_os_020 = echo 4224 > /proc/sys/vm/nr_hugepages %endif %ifdef %{32c750} notes_os_015 = echo 8448 > /proc/sys/vm/nr_hugepages %endif %ifdef %{48c760} notes_os_020 = echo 12672 > /proc/sys/vm/nr_hugepages %endif notes_os_020 = %if '%{benchmark}' eq 'fp' notes_os_025 = The Apache C++ Standard Library V4.2.1 was installed from notes_os_030 = http://stdcxx.apache.org/download.html using: notes_os_035 = gmake BUILDTYPE=8d CONFIG=gcc.config notes_os_040 = %endif notes_os_045 = The following environment varibles were set before the runspec command: notes_os_050 = export HUGETLB_VERBOSE=0 notes_os_055 = export HUGETLB_MORECORE=yes %if '%{benchmark}' eq 'fp' notes_os_060 = export HUGETLB_ELFMAP=RW notes_os_065 = export XLFRTEOPTS=intrinthds=1 %endif #notes_plat_000 = Idle Power Saver was disabled on Advanced System Management. %if '%{benchmark}' eq 'fp' %ifdef %{32c750} notes_plat_000 = Hardware Page Table (HPT) ratio changed as follows by root user on Hardware Management Console (HMC): notes_plat_005 = chsyscfg -m -r prof -i "name=,lpar_name=,hpt_ratio=1:512" %endif %endif ###################################################################### # End of SUT section # If this config file were to be applied to several SUTs, edits would # be needed only ABOVE this point. ###################################################################### %define P7 -qarch=pwr7 -qtune=pwr7 %define LHBDT -B/usr/share/libhugetlbfs/ -tl -Wl,--hugetlbfs-align %define LHLIB -lhugetlbfs ###################################### %define FDPROPTS -O4 -omullX %define FDPRINT %ifdef %{numcores} %define make_j %{numcores} %else %define make_j 4 %endif #====================================================================== # Global Options apply to all runs #====================================================================== # # For just a report generated without a raw file or any runs required # Do a "runspec --reportonly" (also --action=report or --fakereport) # # Just set up the build or run directories, use --buildsetup or --setup # on the runspec command # Sets nobuild as the default - you will need -D to force a compile # nobuild = 1 # -D forces a build (recompile) # -N prohibits a build (recompile) #action = validate # -a tune = base # -T iterations = 1 # -n output_format = text,html,csv # -o #basepeak = 0 reportable = 0 # -l turns off ignore_errors = 1 # -I mean_anyway = 1 allow_extension_override = 1 ext = XL-ppc64 # -e size = ref post_setup = sync parallel_test = 1 # This tells runspec to set enviornment variables of the form # ENV_xxx if found in here. Forced to 0 for reportable runs env_vars = 0 # Control if the embedded md5 string of compiler options is checked for # recompile if the options have changed check_md5 = 0 # Controls if the log file info is echo'd to the screen teeout = yes # If 1, make copy of the config file before modifing with MD5 # signitures of the compile options. backup_config = 0 # verbose -> -v #verbose = 1 to 99, 5 is default, 91 is good for debugging verbose = 1 %ifdef %{make_j} makeflags = -j%{make_j} %endif # We alway use submit to ensure there is a bindprocessor done. # use_submit_for_speed If set, use submit commands for speed as well as rate use_submit_for_speed = 1 # SPEED submit = echo "$command" > run.sh ; numactl --membind=\$SPECCOPYNUM --physcpubind=\$SPECCOPYNUM sh run.sh %ifdef %{48c760} bind = \ 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76 80 84 88 92 96 100 104 108 112 116 120 124 128 132 136 140 144 148 152 156 160 164 168 172 176 180 184 188 \ 1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97 101 105 109 113 117 121 125 129 133 137 141 145 149 153 157 161 165 169 173 177 181 185 189 \ 2 6 10 14 18 22 26 30 34 38 42 46 50 54 58 62 66 70 74 78 82 86 90 94 98 102 106 110 114 118 122 126 130 134 138 142 146 150 154 158 162 166 170 174 178 182 186 190 \ 3 7 11 15 19 23 27 31 35 39 43 47 51 55 59 63 67 71 75 79 83 87 91 95 99 103 107 111 115 119 123 127 131 135 139 143 147 151 155 159 163 167 171 175 179 183 187 191 copies = 192 submit = let "MYCPU=\$BIND"; MYMEM=`printf '%d' \$((\$MYCPU/24))` ; numactl --physcpubind=\$MYCPU --membind=\$MYMEM $command %endif %ifdef %{32c750} bind = \ 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76 80 84 88 92 96 100 104 108 112 116 120 124 \ 1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97 101 105 109 113 117 121 125 \ 2 6 10 14 18 22 26 30 34 38 42 46 50 54 58 62 66 70 74 78 82 86 90 94 98 102 106 110 114 118 122 126 \ 3 7 11 15 19 23 27 31 35 39 43 47 51 55 59 63 67 71 75 79 83 87 91 95 99 103 107 111 115 119 123 127 copies = 128 submit = let "MYCPU=\$BIND"; MYMEM=`printf '%d' \$((\$MYCPU/16))` ; numactl --physcpubind=\$MYCPU --membind=\$MYMEM $command %endif %if defined(%{16c730}) || defined(%{16c740}) || defined(%{16c7R2}) bind = \ 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 \ 1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 \ 2 6 10 14 18 22 26 30 34 38 42 46 50 54 58 62 \ 3 7 11 15 19 23 27 31 35 39 43 47 51 55 59 63 copies = 64 submit = echo "$command" > run.sh; let "MYCPU=\$BIND"; MYMEM=`printf '%d' \$((\$MYCPU/32))` ; numactl --physcpubind=\$MYCPU --membind=\$MYMEM sh run.sh %endif #--prevent log file lines that are too long for vi log_line_width = 130 # Point to where the usr/compiler/OS Flags file is flagsurl000= http://www.spec.org/cpu2006/flags/IBM-Power.20130226.xml flagsurl001 = http://www.spec.org/cpu2006/flags/IBM-Linux-XL.20121024.xml #=============================================================================== # How the compilers are called # The settings used here are for IBM C/C++ and xl Fortran compilers #=============================================================================== default=default=default=default: # IBM Compilers 12.1/14.1 updated to the Dec 2012 PTF CC_PATH = /opt/ibmcmp/vac/12.1/bin CXX_PATH = /opt/ibmcmp/vacpp/12.1/bin XLF_PATH = /opt/ibmcmp/xlf/14.1/bin CC = $(CC_PATH)/xlc -qlanglvl=extc99 CXX = $(CXX_PATH)/xlC FC = $(XLF_PATH)/xlf95 #====================================================================== # Base Level Optimizations # No limit to number of flags, must be same for all of each Language. #====================================================================== default=base=default=default: %ifndef %{debug} %define DEBUG %else %define DEBUG -g %endif # Check for BASEARCH overrides # Default BASEARCH is P7 %define BASEARCH %{P7} %define MODE -q32 %if %{mymode64} %define MODE -q64 %endif # Default mode is -O5 with p7+ %define BASEOPT -O5 %{BASEARCH} %{MODE} %{DEBUG} -qipa=threads #Check for BASEOPT overrides %if %{mybaseopt} == 3 %define BASEOPT -O3 %{BASEARCH} %{MODE} %{DEBUG} -qipa=threads %endif #====================================================================== # INT and FP Portability Flags # no optimizations allowed here #====================================================================== default=default=default=default: %if %{mymode64} CPORTABILITY = -DSPEC_CPU_LP64 CXXPORTABILITY = -DSPEC_CPU_LP64 FPPPORTABILITY = -DSPEC_CPU_LP64 400.perlbench=default=default=default: CPORTABILITY = -DSPEC_CPU_LINUX_PPC -DSPEC_CPU_LP64 403.gcc=peak=default=default: CPORTABILITY = -DSPEC_CPU_LP64 410.bwaves=default=default=default: FPORTABILITY = -qfixed 416.gamess=default=default=default: FPORTABILITY = -qfixed 434.zeusmp=default=default=default: FPORTABILITY = -qfixed FPPPORTABILITY = -DSPEC_CPU_LP64 CPORTABILITY = -DSPEC_CPU_LP64 435.gromacs=default=default=default: FPORTABILITY = -qfixed -qextname 436.cactusADM=default=default=default: FPORTABILITY = -qfixed -qextname CPORTABILITY = -DSPEC_CPU_LP64 FPPPORTABILITY = -DSPEC_CPU_LP64 437.leslie3d=default=default=default: FPORTABILITY = -qfixed CPORTABILITY = -DSPEC_CPU_LP64 FPPPORTABILITY = -DSPEC_CPU_LP64 447.dealII=default=default=default: #srcalt=apache_stdcxx_4_2_1 453.povray=default=default=default: CXXPORTABILITY = -DSPEC_CPU_LP64 454.calculix=default=default=default: FPORTABILITY = -qfixed -qextname 462.libquantum=default=default=default: CPORTABILITY = -DSPEC_CPU_LINUX -DSPEC_CPU_LP64 464.h264ref=default=default=default: CPORTABILITY = -qchars=signed -DSPEC_CPU_LP64 481.wrf=default=default=default: CPORTABILITY = -DNOUNDERSCORE -DSPEC_CPU_LP64 482.sphinx3=default=default=default: CPORTABILITY = -qchars=signed -DSPEC_CPU_LP64 483.xalancbmk=default=default=default: CXXPORTABILITY = -DSPEC_CPU_LINUX -DSPEC_CPU_LP64 %else 400.perlbench=default=default=default: CPORTABILITY = -DSPEC_CPU_LINUX_PPC 462.libquantum=default=default=default: CPORTABILITY = -DSPEC_CPU_LINUX 464.h264ref=default=default=default: CPORTABILITY = -qchars=signed 483.xalancbmk=default=default=default: CXXPORTABILITY = -DSPEC_CPU_LINUX 410.bwaves=default=default=default: FPORTABILITY = -qfixed 416.gamess=default=default=default: FPORTABILITY = -qfixed 434.zeusmp=default=default=default: FPORTABILITY = -qfixed 435.gromacs=default=default=default: FPORTABILITY = -qfixed -qextname 436.cactusADM=default=default=default: FPORTABILITY = -qfixed -qextname 437.leslie3d=default=default=default: FPORTABILITY = -qfixed 447.dealII=default=default=default: #srcalt=apache_stdcxx_4_2_1 454.calculix=default=default=default: FPORTABILITY = -qfixed -qextname 481.wrf=default=default=default: CPORTABILITY = -DNOUNDERSCORE 482.sphinx3=default=default=default: CPORTABILITY = -qchars=signed %endif # mymode64 #====================================================================== # INT and FP Base Flags #====================================================================== int=base=default=default: COPTIMIZE = %{BASEOPT} -qalias=noansi -qalloca CXXOPTIMIZE = %{BASEOPT} -qrtti EXTRA_CLIBS = %{LHLIB} EXTRA_CXXLIBS = -lsmartheap EXTRA_LDFLAGS = %{MODE} fp=base=default=default: COPTIMIZE = %{BASEOPT} %{LHBDT} #CXXOPTIMIZE = %{BASEOPT} -qrtti -qstaticlink CXXOPTIMIZE = %{BASEOPT} -qrtti #EXTRA_CXXLIBS = -Wl,--whole-archive /usr/lib/libhugetlbfs.a -Wl,--no-whole-archive EXTRA_CXXLIBS = %{LHBDT} FOPTIMIZE = %{BASEOPT} -qalias=nostd %{LHBDT} EXTRA_LDFLAGS = %{MODE} #====================================================================== # # Default peak flags. # #====================================================================== default=peak=default=default: %define O2 -O2 %{BASEARCH} %{DEBUG} %define O3 -O3 %{BASEARCH} %{DEBUG} %define O4 -O4 %{BASEARCH} %{DEBUG} -qipa=threads %define O5 -O5 %{BASEARCH} %{DEBUG} -qipa=threads #====================================================================== # INT peak Flags #====================================================================== int=peak=default=default: 400.perlbench=peak=default=default: # C only Benchmark %ifdef %{debug_copies} copies = %{debug_copies} %endif #fdpr 5503 failed with -O4 -vrox OPTIMIZE = %{O4} -qalias=noansi -qipa=level=2 EXTRA_LIBS = -lsmartheap LDCFLAGS = -Wl,-q PASS1_CFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_CFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 fdo_pre2 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr -imullX ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run2 = $command fdo_post2 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O4 -omullX ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} CPORTABILITY = -DSPEC_CPU_LINUX_PPC -DSPEC_CPU_LP64 COPTIMIZE = -q64 %else CPORTABILITY = -DSPEC_CPU_LINUX_PPC %endif notes_peak_fdpr_400_000= Post-Link optimization tool used for: notes_peak_fdpr_400_005= 400.perlbench notes_peak_fdpr_400_010= with options -O4 -omullX for optimization phase, notes_peak_fdpr_400_015= and -imullX for instrumentation phase 401.bzip2=peak=default=default: # C only Benchmark %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O3} EXTRA_LIBS = %{LHLIB} LDCFLAGS = -Wl,-q PASS1_CFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_CFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 fdo_pre2 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr %{FDPRINT} ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run2 = $command fdo_post2 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O4 -vrox ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} COPTIMIZE = -q64 %endif notes_peak_fdpr_401_000= 401.bzip2 notes_peak_fdpr_401_005= with options -O4 -vrox 403.gcc=peak=default=default: # C only Benchmark %if defined(%{16c730}) || defined(%{16c740}) || defined(%{16c7R2}) copies = 32 submit = echo "$command" > run.sh; let "CHECK=\$SPECCOPYNUM%2" ; if [[ \$CHECK -eq 0 ]]; then let "MYCPU=2*\$SPECCOPYNUM" ; fi; if [[ \$CHECK -eq 1 ]]; then let "MYCPU=2*\$SPECCOPYNUM-1" ; fi; MYMEM=`printf '%d' \$((\$MYCPU/32))` ; numactl --physcpubind=\$MYCPU --membind=\$MYMEM sh run.sh %endif %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O4} -qalloca EXTRA_LIBS = %{LHLIB} LDCFLAGS = -Wl,-q PASS1_CFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_CFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 fdo_pre2 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr %{FDPRINT} ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run2 = $command fdo_post2 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O4 -nodp -rtb ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} CPORTABILITY = -DSPEC_CPU_LP64 COPTIMIZE = -q64 %endif notes_peak_fdpr_403_000= 403.gcc notes_peak_fdpr_403_005= with options -O4 -nodp -rtb 429.mcf=peak=default=default: # C only Benchmark %if defined(%{16c730}) || defined(%{16c740}) || defined(%{16c7R2}) copies = 32 submit = echo "$command" > run.sh; let "CHECK=\$SPECCOPYNUM%2" ; if [[ \$CHECK -eq 0 ]]; then let "MYCPU=2*\$SPECCOPYNUM" ; fi; if [[ \$CHECK -eq 1 ]]; then let "MYCPU=2*\$SPECCOPYNUM-1" ; fi; MYMEM=`printf '%d' \$((\$MYCPU/32))` ; numactl --physcpubind=\$MYCPU --membind=\$MYMEM sh run.sh %endif %ifdef %{32c750} copies = 64 %endif %ifdef %{48c760} copies = 48 %endif %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O5} EXTRA_LIBS = %{LHLIB} LDCFLAGS = -Wl,-q fdo_pre1 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr %{FDPRINT} ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run1 = $command fdo_post1 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O3 ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} COPTIMIZE = -q64 %endif notes_peak_fdpr_429_000= 429.mcf 445.gobmk 458.sjeng 473.astar notes_peak_fdpr_429_005= with options -O3 445.gobmk=peak=default=default: %ifdef %{debug_copies} copies = %{debug_copies} %endif # C only Benchmark OPTIMIZE = %{O4} EXTRA_LIBS = %{LHLIB} LDCFLAGS = -Wl,-q PASS1_CFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_CFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 fdo_pre2 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr %{FDPRINT} ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run2 = $command fdo_post2 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O3 ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} COPTIMIZE = -q64 %endif 456.hmmer=peak=default=default: %ifdef %{debug_copies} copies = %{debug_copies} %endif # C only Benchmark OPTIMIZE = %{O5} -qsimd -qassert=refalign -qipa=inline=threshold=2888 -qipa=inline=limit=11880 EXTRA_LIBS = %{LHLIB} LDCFLAGS = -Wl,-q fdo_pre1 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr %{FDPRINT} ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run1 = $command fdo_post1 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O4 -nodp -m power7 ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} COPTIMIZE = -q64 %endif 458.sjeng=peak=default=default: %ifdef %{debug_copies} copies = %{debug_copies} %endif # C only Benchmark OPTIMIZE = %{O5} EXTRA_LIBS = %{LHLIB} LDCFLAGS = -Wl,-q PASS1_CFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_CFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 fdo_pre2 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr %{FDPRINT} ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run2 = $command fdo_post2 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O3 ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} COPTIMIZE = -q64 %endif 462.libquantum=peak=default=default: # C only Benchmark %if defined(%{16c730}) || defined(%{16c740}) copies=64 %endif %ifdef %{debug_copies} copies = %{debug_copies} %endif %define 64 OPTIMIZE = %{O5} EXTRA_LIBS = %{LHLIB} LDCFLAGS = -Wl,-q PASS1_CFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_CFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 fdo_pre2 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr %{FDPRINT} ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run2 = $command fdo_post2 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O4 -vrox -nodp ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} COPTIMIZE = -q64 %endif %undef 64 notes_peak_fdpr_462_000= 462.libquantum notes_peak_fdpr_462_005= with options -O4 -vrox -nodp 464.h264ref=peak=default=default: # C only Benchmark #%define 64 %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O5} EXTRA_LIBS = %{LHLIB} LDCFLAGS = -Wl,-q PASS1_CFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_CFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 fdo_pre2 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run2 = $command fdo_post2 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O4 -vrox -nodp -rtb ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} COPTIMIZE = -q64 %endif #%undef 64 notes_peak_fdpr_464_000= 464.h264ref notes_peak_fdpr_464_005= with options -O4 -vrox -nodp -rtb 471.omnetpp=peak=default=default: # C++ Benchmark %ifdef %{32c750} copies = 32 %endif %ifdef %{48c760} copies = 48 %endif %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O5} -qrtti EXTRA_LIBS = -lsmartheap LDCXXFLAGS = -Wl,-q PASS1_CXXFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_CXXFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 fdo_pre2 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run2 = $command fdo_post2 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O3 -lu -1 -nodp -sdp 9 ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} CXXOPTIMIZE = -q64 %endif notes_peak_fdpr_471_000= 471.omnetpp notes_peak_fdpr_471_005= with options -O3 -lu -1 -nodp -sdp 9 473.astar=peak=default=default: # C++ Benchmark %ifdef %{32c750} copies = 96 %endif %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O4} EXTRA_LIBS = %{LHLIB} -lsmartheap LDCXXFLAGS = -Wl,-q PASS1_CXXFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_CXXFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 fdo_pre2 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr %{FDPRINT} ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run2 = $command fdo_post2 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O3 ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} CXXOPTIMIZE = -q64 %endif 483.xalancbmk=peak=default=default: # C++ Benchmark %ifdef %{48c760} copies = 96 %endif %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O4} -qipa=partition=large EXTRA_LIBS = -lsmartheap LDCXXFLAGS = -Wl,-q fdo_pre1 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr %{FDPRINT} ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run1 = $command fdo_post1 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O3 -m power7 ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} CXXOPTIMIZE = -q64 %endif notes_peak_fdpr_483_000= 483.xalancbmk notes_peak_fdpr_483_005= with options -O3 -m power7 999.specrand=peak=default=default: # C only Benchmark - Must run, but not reported unless fails %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O5} EXTRA_LIBS = %{LHLIB} PASS1_CFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_CFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 #====================================================================== # FP peak Flags #====================================================================== fp=peak=default=default: 410.bwaves=peak=default=default: # Fortran Benchmark #got 3% gain with -qsmallstack in single-threaded. %if defined(%{16c730}) || defined(%{16c740}) || defined(%{16c7R2}) copies = 16 submit = echo "$command" > run.sh; let "MYCPU=4*\$SPECCOPYNUM"; MYMEM=`printf '%d' \$((\$MYCPU/32))` ; numactl --physcpubind=\$MYCPU --membind=\$MYMEM sh run.sh %endif %ifdef %{32c750} copies = 32 %endif %ifdef %{48c760} copies = 48 %endif %ifdef %{debug_copies} copies = %{debug_copies} %endif %define 64 OPTIMIZE = %{O4} -qsmallstack=dynlenonheap EXTRA_LIBS = %{LHLIB} PASS1_FFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_FFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 %ifdef %{64} FOPTIMIZE = -q64 %endif %undef 64 416.gamess=peak=default=default: # Fortran Benchmark %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O5} -qalias=nostd EXTRA_LIBS = %{LHLIB} PASS1_FFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_FFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 %ifdef %{64} FOPTIMIZE = -q64 %endif 433.milc=peak=default=default: # C only Benchmark %if defined(%{16c730}) || defined(%{16c740}) || defined(%{16c7R2}) copies = 16 submit = echo "$command" > run.sh; let "MYCPU=4*\$SPECCOPYNUM"; MYMEM=`printf '%d' \$((\$MYCPU/32))` ; numactl --physcpubind=\$MYCPU --membind=\$MYMEM sh run.sh %endif %ifdef %{32c750} copies = 32 %endif %ifdef %{48c760} copies = 48 %endif %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O5} EXTRA_LIBS = %{LHLIB} LDCFLAGS = -Wl,-q fdo_pre1 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr %{FDPRINT} ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run1 = $command fdo_post1 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O4 -nodp ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} COPTIMIZE = -q64 %endif notes_peak_fdpr_433_000= Post-Link optimization tool used for: notes_peak_fdpr_433_005= 433.milc 435.gromacs 450.soplex 482.sphinx3 notes_peak_fdpr_433_010 = with options -O4 -nodp 434.zeusmp=peak=default=default: # Fortran Benchmark %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O3} -qxlf90=nosignedzero EXTRA_LIBS = %{LHBDT} PASS1_FFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_FFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 LDFFLAGS = -Wl,-q fdo_pre2 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr %{FDPRINT} ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run2 = $command fdo_post2 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O4 -vrox -nodp ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} FOPTIMIZE = -q64 %endif notes_peak_fdpr_434_000= 434.zeusmp notes_peak_fdpr_434_005 = with options -O4 -vrox -nodp 435.gromacs=peak=default=default: # Fortran & C Benchmark %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O4} -qsimd EXTRA_LIBS = %{LHLIB} LDCFLAGS = -Wl,-q LDFFLAGS = -Wl,-q PASS1_FFLAGS = -qpdf1 PASS1_CFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_FFLAGS = -qpdf2 PASS2_CFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 fdo_pre2 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr %{FDPRINT} ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run2 = $command fdo_post2 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O4 -nodp ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} COPTIMIZE = -q64 FOPTIMIZE = -q64 %endif 436.cactusADM=peak=default=default: # Fortran & C Benchmark %if defined(%{16c730}) || defined(%{16c740}) || defined(%{16c7R2}) copies = 32 submit = echo "$command" > run.sh; let "CHECK=\$SPECCOPYNUM%2" ; if [[ \$CHECK -eq 0 ]]; then let "MYCPU=2*\$SPECCOPYNUM" ; fi; if [[ \$CHECK -eq 1 ]]; then let "MYCPU=2*\$SPECCOPYNUM-1" ; fi; MYMEM=`printf '%d' \$((\$MYCPU/32))` ; numactl --physcpubind=\$MYCPU --membind=\$MYMEM sh run.sh %endif %ifdef %{32c750} copies = 32 %endif %ifdef %{48c760} copies = 48 %endif %ifdef %{debug_copies} copies = %{debug_copies} %endif %define 64 OPTIMIZE = %{O4} -qsimd -qnostrict FOPTIMIZE = -qsmallstack=dynlenonheap -qalias=nostd EXTRA_LIBS = %{LHLIB} %ifdef %{64} FPORTABILITY = -qfixed -qextname FPPPORTABILITY = -DSPEC_CPU_LP64 CPORTABILITY = -DSPEC_CPU_LP64 FOPTIMIZE = -q64 COPTIMIZE = -q64 %else FPORTABILITY = -qfixed -qextname %endif %undef 64 437.leslie3d=peak=default=default: # Fortran Benchmark %if defined(%{16c730}) || defined(%{16c740}) || defined(%{16c7R2}) copies = 16 submit = echo "$command" > run.sh; let "MYCPU=4*\$SPECCOPYNUM"; MYMEM=`printf '%d' \$((\$MYCPU/32))` ; numactl --physcpubind=\$MYCPU --membind=\$MYMEM sh run.sh %endif %ifdef %{32c750} copies = 32 %endif %ifdef %{48c760} copies = 48 %endif %ifdef %{debug_copies} copies = %{debug_copies} %endif %define 64 OPTIMIZE = %{O5} EXTRA_LIBS = %{LHBDT} LDFFLAGS = -Wl,-q fdo_pre1 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr %{FDPRINT} ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run1 = $command fdo_post1 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O3 -lu -1 -nodp -sdp 9 ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} FOPTIMIZE = -q64 %endif %undef 64 notes_peak_fdpr_437_000 = 437.leslie3d notes_peak_fdpr_437_005 = with options -O3 -lu -1 -nodp -sdp 9 444.namd=peak=default=default: # C++ Benchmark %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O5} EXTRA_LIBS = %{LHLIB} LDCXXFLAGS = -Wl,-q PASS1_CXXFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_CXXFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 fdo_pre2 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr %{FDPRINT} ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run2 = $command fdo_post2 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O3 -lu -1 -nodp -sdp 9 ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} CXXOPTIMIZE = -q64 %endif notes_peak_fdpr_444_000 = 444.namd notes_peak_fdpr_444_005 = with options -O3 -lu -1 -nodp -sdp 9 447.dealII=peak=default=default: # C++ Benchmark %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O4} -qrtti EXTRA_CXXFLAGS = -qcpp_stdinc=/opt/stdcxx421/include/ansi:/opt/stdcxx421/include:/opt/ibmcmp/vacpp/12.1/include EXTRA_CXXLIBS = -L/opt/stdcxx421/lib -R/opt/stdcxx421/lib -lstd8d EXTRA_LIBS = -lsmartheap %ifdef %{64} CXXOPTIMIZE = -q64 %endif 450.soplex=peak=default=default: # C++ Benchmark %if defined(%{16c730}) || defined(%{16c740}) || defined(%{16c7R2}) copies = 32 %endif %ifdef %{48c760} copies = 48 %endif %ifdef %{debug_copies} copies = %{debug_copies} %endif %define 64 OPTIMIZE = %{O3} EXTRA_LIBS = %{LHLIB} LDCXXFLAGS = -Wl,-q PASS1_CXXFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_CXXFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 fdo_pre2 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr %{FDPRINT} ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run2 = $command fdo_post2 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O4 -nodp ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} CXXOPTIMIZE = -q64 %endif %undef 64 notes_peak_fdpr_450_000 = 450.soplex notes_peak_fdpr_450_005 = with options -O4 -nodp 453.povray=peak=default=default: # C++ Benchmark %ifdef %{debug_copies} copies = %{debug_copies} %endif %define 64 OPTIMIZE = %{O4} -qsimd EXTRA_LIBS = -lsmartheap64 LDCXXFLAGS = -Wl,-q PASS1_CXXFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_CXXFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 #FDPR XL11 + 5.5.0.2 brokes it #FDPR XL11 + 5.5.0.3 looping with ref input #fdo_pre2 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr %{FDPRINT} ${baseexe}; cp ${baseexe}.instr ${baseexe} #fdo_run2 = $command #fdo_post2 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O3 ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} CXXPORTABILITY = -DSPEC_CPU_LP64 CXXOPTIMIZE = -q64 %endif %undef 64 454.calculix=peak=default=default: # Fortran & C Benchmark %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O5} EXTRA_LIBS = %{LHBDT} ## fdpr 5503 failed when using with LHBDT #LDCFLAGS = -Wl,-q #LDFFLAGS = -Wl,-q PASS1_FFLAGS = -qpdf1 PASS1_CFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_FFLAGS = -qpdf2 PASS2_CFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 #fdo_pre2 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr ${baseexe}; cp ${baseexe}.instr ${baseexe} #fdo_run2 = $command #fdo_post2 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O3 ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} FOPTIMIZE = -q64 COPTIMIZE = -q64 %endif 459.GemsFDTD=peak=default=default: # Fortran Benchmark %if defined(%{16c730}) || defined(%{16c740}) || defined(%{16c7R2}) copies = 16 submit = echo "$command" > run.sh; let "MYCPU=4*\$SPECCOPYNUM"; MYMEM=`printf '%d' \$((\$MYCPU/32))` ; numactl --physcpubind=\$MYCPU --membind=\$MYMEM sh run.sh %endif %ifdef %{48c760} copies = 48 %endif %ifdef %{32c750} copies = 64 %endif %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O4} -qsimd EXTRA_LIBS = %{LHBDT} %ifdef %{64} FOPTIMIZE = -q64 %endif 465.tonto=peak=default=default: # Fortran Benchmark %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O5} -qsimd # SmartHeap version 9 is not supported with XL11 EXTRA_LIBS = %{LHLIB} PASS1_FFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_FFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 LDFFLAGS = -Wl,-q fdo_pre2 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run2 = $command fdo_post2 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O4 ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} FOPTIMIZE = -q64 %endif notes_peak_fdpr_465_000 = 465.tonto notes_peak_fdpr_465_005 = with options -O4 470.lbm=peak=default=default: # C Benchmark %ifdef %{debug_copies} copies = %{debug_copies} %endif basepeak=yes #%if defined(%{16c730}) || defined(%{16c740}) || defined(%{16c7R2}) #copies = 32 #submit = echo "$command" > run.sh; let "CHECK=\$SPECCOPYNUM%2" ; if [[ \$CHECK -eq 0 ]]; then let "MYCPU=2*\$SPECCOPYNUM" ; fi; if [[ \$CHECK -eq 1 ]]; then let "MYCPU=2*\$SPECCOPYNUM-1" ; fi; MYMEM=`printf '%d' \$((\$MYCPU/32))` ; numactl --physcpubind=\$MYCPU --membind=\$MYMEM sh run.sh #%endif #%if %{numcores} == 128 #copies = 128 #submit = echo "$command" > run.sh; let "MYCPU=4*\$SPECCOPYNUM"; MYMEM=`printf '%d' \$((\$MYCPU/32))` ; numactl --physcpubind=\$MYCPU --membind=\$MYMEM sh run.sh #%endif #OPTIMIZE = %{O5} #EXTRA_LIBS = %{LHBDT} #PASS1_CFLAGS = -qpdf1 #PASS1_LDFLAGS = -qpdf1 #PASS2_CFLAGS = -qpdf2 #PASS2_LDFLAGS = -qpdf2 #%ifdef %{64} #COPTIMIZE = -q64 #%endif 481.wrf=peak=default=default: # Fortran & C-for netCDF Benchmark #with or w/o smallstack make no diff # -O3 is better than O5 %ifdef %{48c760} copies = 96 %endif %ifdef %{debug_copies} copies = %{debug_copies} %endif %define 64 OPTIMIZE = %{O3} EXTRA_LIBS = %{LHLIB} %ifdef %{64} COPTIMIZE = -q64 FOPTIMIZE = -q64 %endif %undef 64 482.sphinx3=peak=default=default: # C only Benchmark %if defined(%{16c730}) || defined(%{16c740}) || defined(%{16c7R2}) copies = 16 submit = echo "$command" > run.sh; let "MYCPU=4*\$SPECCOPYNUM"; MYMEM=`printf '%d' \$((\$MYCPU/32))` ; numactl --physcpubind=\$MYCPU --membind=\$MYMEM sh run.sh %endif %ifdef %{32c750} copies = 32 %endif %ifdef %{48c760} copies = 48 %endif %ifdef %{debug_copies} copies = %{debug_copies} %endif OPTIMIZE = %{O4} EXTRA_LIBS = %{LHLIB} LDCFLAGS = -Wl,-q PASS1_CFLAGS = -qpdf1 PASS1_LDFLAGS = -qpdf1 PASS2_CFLAGS = -qpdf2 PASS2_LDFLAGS = -qpdf2 fdo_pre2 = cp ${baseexe} ${baseexe}.orig ; /opt/bin/fdprpro -a instr %{FDPRINT} ${baseexe}; cp ${baseexe}.instr ${baseexe} fdo_run2 = $command fdo_post2 = cp ${baseexe}.orig ${baseexe} ; /opt/bin/fdprpro -a opt -f ${baseexe}.nprof -O4 -nodp ${baseexe} ; cp ${baseexe}.fdpr ${baseexe} %ifdef %{64} COPTIMIZE = -q64 %endif notes_peak_fdpr_482_000 = 482.sphinx3 notes_peak_fdpr_482_005 = with options -O4 -nodp 998.specrand=peak=default=default: # C only Benchmark - Must run, but not reported unless fails OPTIMIZE = %{O5} 434.zeusmp: # The following setting was inserted automatically as a result of # post-run basepeak application. basepeak = 1