# Invocation command line: # /root/work/cpu2006/bin/runspec -c amd1002li-speed-revA.cfg -T all int --nopreenv --note-preenv # output_root was not used for this run ############################################################################ # AMD SPEC CPU2006 V1.1 Speed Configuration File for 64-bit Linux # # CPU2006 Version 1.1 # Compiler name/version: Open64 4.2.3.2 # Operating system version: 64-bit Linux # Hardware: AMD Lisbon # FP Base Pointer Size: 64-bit only # FP Peak Pointer Size: 32/64-bit # INT Base Pointer Size: 32/64-bit # INT Peak Pointer Size: 32/64-bit # Auto Parallization: Yes # # Important! Please run with your stack size set to 'unlimited'. # Failure to do so may cause 483.xalancbmk to get a stack overflow during # execution. # Using csh: unlimit # Using bash: ulimit -s unlimited # # Your LD_LIBRARY_PATH must be set to the location of the dependent # runtime libraries. (Though this should be taken care of by the # preENV_LD_LIBRARY_PATH found in the AMD1002li-speed-revX.inc file.) # This line should be commented out during builds. # ##################################################################### ext = amd1002mc-speed-revA ignore_errors = no tune = base,peak output_format = all flagsurl000 = http://www.spec.org/cpu2006/flags/x86-open64-423-flags-speed-revA.20101207.xml $[top]/amd-platform-speed-revA.xml size = test,train,ref check_md5 = yes reportable = yes env_vars = no teeout = yes mean_anyway = yes teerunout = yes post_setup = sync use_submit_for_speed = 1 #################################################################### # Include file containing the commonly changed fields #################################################################### %define inc_file_name AMD1002li-speed-revA.inc #include: %{inc_file_name} # ----- Begin inclusion of 'AMD1002li-speed-revA.inc' ############################################################################ ##################################################################### # Paths ##################################################################### # Set location of run time libraries for runs. preENV_LD_LIBRARY_PATH = $[top]/amd1002-speed-libs-revA/64:$[top]/amd1002-speed-libs-revA/32 # Set this path to your SmartHeap Library for builds. SMARTHEAP_DIR = /root/work/libraries/SmartHeap-8.1/lib ##################################################################### # Machine Specific Settings ##################################################################### # If your machine orders nodes sequentially, you have 8GB/socket # memory, and your system has huge pages configured, most settings # will be correctly configured based on the below two settings. See # README.amd1002li-speed for more information. # How many sockets does your test system have? (1 or 2) %define num_sockets 2 # How many cores per socket does your system have? (4 or 6) %define cores_per_socket 4 ######################################################################## # You should not need to change anything in this section unless you have # a non-standard configuration. See README.amd1002li-speed for more # information. ######################################################################## %if ('%{cores_per_socket}' eq '6') && ('%{num_sockets}' eq '1') % define num_system_huge_pages 1000 % define num_cores 6 %elif ('%{cores_per_socket}' eq '6') && ('%{num_sockets}' eq '2') % define num_system_huge_pages 2000 % define num_cores 12 %elif ('%{cores_per_socket}' eq '4') && ('%{num_sockets}' eq '1') % define num_system_huge_pages 1000 % define num_cores 4 %elif ('%{cores_per_socket}' eq '4') && ('%{num_sockets}' eq '2') % define num_system_huge_pages 2000 % define num_cores 8 %endif #################################################################### # Tester information #################################################################### license_num = 49 prepared_by = AMD Performance Lab tester = Advanced Micro Devices test_sponsor = Advanced Micro Devices hw_vendor = Tyan hw_model000 = Tyan YR190B8228, hw_model001 = AMD Opteron 4122 #################################################################### # Hardware information #################################################################### hw_avail = Aug-2010 hw_cpu_name = AMD Opteron 4122 hw_cpu_mhz = 2200 hw_fpu = Integrated hw_nchips = 2 hw_ncores = 8 hw_ncoresperchip = 4 hw_nthreadspercore = 1 hw_ncpuorder = 1,2 chips hw_pcache = 64 KB I + 64 KB D on chip per core hw_scache = 512 KB I+D on chip per core hw_tcache = 6 MB I+D on chip per chip hw_ocache = None hw_disk000 = 1 x 128 GB SATA SSD hw_disk001 = Crucial RealSSD C300 CTFDDAC128MAG-1G1 hw_memory = 32 GB (4 x 8 GB 2Rx4 PC3-10600R-9, ECC) hw_other = None sw_file = ext3 #sw_os000 = Red Hat Enterprise Linux Server release 5.4, #sw_os001 = Kernel 2.6.18-164.9.1.el5 sw_os000 = SUSE Linux Enterprise Server 11 (x86_64), sw_os001 = Kernel 2.6.27.19-5-default sw_state = Run level 3 (Full multiuser with network) ##################################################################### # Notes ##################################################################### notes_025 =The x86 Open64 Compiler Suite is only available from (and supported by) AMD at notes_030 =http://developer.amd.com/cpu/open64 notes_submit_000 ='numactl' was used to bind copies to the cores. notes_submit_005 =See the configuration file for details. notes_os_000 ='ulimit -s unlimited' was used to set environment stack size notes_os_005 ='ulimit -l 2097152' was used to set environment locked pages in memory limit notes_os_010 = notes_os_015 =Set vm/nr_hugepages=2000 in /etc/sysctl.conf notes_os_020 =mount -t hugetlbfs nodev /mnt/hugepages notes_os_025 = # SLES notes_os_030 =powersave -f was used to set the CPU frequency to its maximum. # RHEL #notes_os_030 =cpuspeed stop was used to set the CPU frequency to its maximum. # SLES 11 and later notes_os_035 = notes_os_040 =Binaries were compiled on SLES10 SP2 with binutils 2.18 # ---- End inclusion of '/root/work/cpu2006/config/AMD1002li-speed-revA.inc' #################################################### # Auto-configured settings based on your .inc file # #################################################### preENV_O64_OMP_SPIN_USER_LOCK = true %if '%{cores_per_socket}' eq '6' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3,4,5 bash run.sh % if '%{num_sockets}' eq '1' # Settings for 1P 6 core preENV_O64_OMP_AFFINITY_MAP = 0,1,2,3,4,5 % elif '%{num_sockets}' eq '2' # Settings for 2P 6 core preENV_O64_OMP_AFFINITY_MAP = 0,1,2,3,4,5,6,7,8,9,10,11 % else #num_sockets % error Please define "num_sockets" to either '1' or '2' in %{inc_file_name}. This error was generated % endif #num_sockets %elif '%{cores_per_socket}' eq '4' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3 bash run.sh % if '%{num_sockets}' eq '1' # Settings for 1P 4 core preENV_O64_OMP_AFFINITY_MAP = 0,1,2,3 % elif '%{num_sockets}' eq '2' # Settings for 2P 4 core preENV_O64_OMP_AFFINITY_MAP = 0,1,2,3,4,5,6,7 % else #num_sockets % error Please define "num_sockets" to either '1' or '2' in %{inc_file_name}. This error was generated % endif #num_sockets %else #cores_per_socket % error Please define "cores_per_socket" to either '4' or '6' in %{inc_file_name}. This error was generated %endif #cores_per_socket ################################ # End auto-configured settings # ################################ ########################## Software Info ############################ fp=default=default=default: sw_base_ptrsize = 64-bit sw_peak_ptrsize = 32/64-bit sw_avail = May-2010 sw_compiler = x86 Open64 4.2.3.2 Compiler Suite (from AMD) #sw_other1 = binutils 2.18 int=default=default=default: sw_base_ptrsize = 32/64-bit sw_peak_ptrsize = 32/64-bit sw_avail = May-2010 sw_compiler = x86 Open64 4.2.3.2 Compiler Suite (from AMD) sw_other = SmartHeap 8.1 32-bit Library for Linux #sw_other2 = binutils 2.18 default=default=default=default: ##################################################################### # Compiler selection # # CC = opencc -march=barcelona CXX = openCC -march=barcelona FC = openf95 -march=barcelona ##################################################################### # Portability for default 64-bit code generation ##################################################################### default=default=default: PORTABILITY = -DSPEC_CPU_LP64 ##################################################################### # INT Portability ##################################################################### 400.perlbench=default=default: CPORTABILITY = -DSPEC_CPU_LINUX_X64 403.gcc=peak=default: PORTABILITY = 429.mcf=peak=default: PORTABILITY = 462.libquantum=default=default: CPORTABILITY = -DSPEC_CPU_LINUX 471.omnetpp=default=default: # Needed to avoid -DSPEC_CPU_LP64 on -m32 C++ base codes PORTABILITY = 473.astar=default=default: # Needed to avoid -DSPEC_CPU_LP64 on -m32 C++ base codes PORTABILITY = 483.xalancbmk=default=default: CXXPORTABILITY= -DSPEC_CPU_LINUX # Needed to avoid -DSPEC_CPU_LP64 on -m32 C++ base codes PORTABILITY = ################################################################# # # # SPECint Tuning & Notes # # # ################################################################# int=base: COPTIMIZE = -Ofast -apo -CG:local_sched_alg=1 -HP:bdt=2m:heap=2m,limit=450 -LNO:parallel_overhead=10000 CXXOPTIMIZE = -Ofast -m32 -INLINE:aggressive=on -CG:cmp_peep=on EXTRA_CXXLIBS = -L$(SMARTHEAP_DIR) -lsmartheap sw_base_ptrsize = 32/64-bit sw_peak_ptrsize = 32/64-bit %if '%{cores_per_socket}' eq '6' % if '%{num_sockets}' eq '1' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3,4,5 bash run.sh % elif '%{num_sockets}' eq '2' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3,4,5,6,7,8,9,10,11 bash run.sh % endif #num_sockets %elif '%{cores_per_socket}' eq '4' % if '%{num_sockets}' eq '1' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3 bash run.sh % elif '%{num_sockets}' eq '2' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3,4,5,6,7 bash run.sh % endif #num_sockets %endif #cores_per_socket ##################################################################### # INT Peak Tuning ##################################################################### int=peak=default: COPTIMIZE = -Ofast -apo -CG:local_sched_alg=1 -HP:bdt=2m:heap=2m,limit=450 -LNO:parallel_overhead=10000 CXXOPTIMIZE = -Ofast -m32 -HP:bdt=2m:heap=2m PASS1_CFLAGS = -fb_create fbdata PASS1_CXXFLAGS = -fb_create fbdata PASS1_LDFLAGS = -fb_create fbdata PASS2_CFLAGS = -fb_opt fbdata PASS2_CXXFLAGS = -fb_opt fbdata PASS2_LDFLAGS = -fb_opt fbdata 400.perlbench=peak=default: COPTIMIZE = -Ofast -IPA:plimit=20000 -LNO:opt=0 -OPT:unroll_times_max=8:unroll_size=256:unroll_level=2:keep_ext=on -WOPT:if_conv=0 -CG:local_sched_alg=1:unroll_fb_req=on -HP:bdt=2m:heap=2m 401.bzip2=peak=default: COPTIMIZE = -O3 -OPT:alias=disjoint:goto=off -CG:local_sched_alg=1 -HP:bdt=2m:heap=2m 403.gcc=peak=default: COPTIMIZE = -Ofast -LNO:trip_count=256:prefetch_ahead=10 -CG:cmp_peep=on -m32 -HP:bdt=2m:heap=2m -GRA:unspill=on -IPA:small_pu=200 429.mcf=peak=default: COPTIMIZE = -O3 -ipa -INLINE:aggressive=on -CG:gcm=off -GRA:prioritize_by_density=on -m32 -HP:bdt=2m:heap=2m feedback = 0 445.gobmk=peak=default: COPTIMIZE = -O3 -OPT:alias=restrict:unroll_times_max=8:unroll_size=256:unroll_level=2:keep_ext=on -ipa -IPA:plimit=750:min_hotness=300:pu_reorder=1 -LNO:prefetch=1:ignore_feedback=off -CG:p2align=on:unroll_fb_req=on -HP:bdt=2m:heap=2m 456.hmmer=peak=default: COPTIMIZE = -Ofast -LNO:prefetch=0 -OPT:alias=disjoint:unroll_times_max=8:unroll_size=256:unroll_level=2:keep_ext=on -CG:local_sched_alg=1:cflow=0:push_pop_int_saved_regs=off:cmp_peep=on -HP:bdt=2m:heap=2m 458.sjeng=peak=default: COPTIMIZE = -O3 -ipa -LNO:ignore_feedback=off:full_unroll=10:fusion=0:fission=2 -IPA:pu_reorder=2:min_hotness=32 -CG:ptr_load_use=0 -OPT:unroll_times_max=8 -INLINE:aggressive=on -HP:bdt=2m:heap=2m 462.libquantum=peak=default: COPTIMIZE = -Ofast -apo -LNO:pf2=0 -CG:gcm=off:use_prefetchnta=on:cmp_peep=on -WOPT:aggstr=0 -OPT:alias=disjoint -INLINE:aggressive=on -IPA:space=1000:plimit=20000 -mso feedback = 0 %if '%{cores_per_socket}' eq '6' % if '%{num_sockets}' eq '1' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3,4,5 bash run.sh % elif '%{num_sockets}' eq '2' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3,4,5,6,7,8,9,10,11 bash run.sh % endif #num_sockets %elif '%{cores_per_socket}' eq '4' % if '%{num_sockets}' eq '1' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3 bash run.sh % elif '%{num_sockets}' eq '2' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3,4,5,6,7 bash run.sh % endif #num_sockets %endif #cores_per_socket 464.h264ref=peak=default: COPTIMIZE = -O3 -IPA:plimit=20000 -OPT:alias=disjoint -LNO:prefetch=0 -CG:ptr_load_use=0:push_pop_int_saved_regs=off -HP:bdt=2m:heap=2m 471.omnetpp=peak=default: CXXOPTIMIZE = -Ofast -CG:gcm=off -INLINE:aggressive=on -WOPT:if_conv=0 -m32 -HP:bdt=2m:heap=2m feedback = 0 473.astar=peak=default: CXXOPTIMIZE = -Ofast -TENV:frame_pointer=off -WOPT:if_conv=0 -GRA:optimize_boundary=on -OPT:alias=disjoint -INLINE:aggressive=on -IPA:small_pu=3000:plimit=3000 -m32 -HP:bdt=2m:heap=2m 483.xalancbmk=peak=default: CXXOPTIMIZE = -Ofast -INLINE:aggressive=on -m32 -CG:cmp_peep=on -GRA:unspill=on -TENV:frame_pointer=off -fno-emit-exceptions EXTRA_CXXLIBS = -L$(SMARTHEAP_DIR) -lsmartheap feedback = 0 ##################################################################### # FP Portability ##################################################################### 436.cactusADM=default=default: FPORTABILITY = -fno-second-underscore 447.dealII=peak=default: PORTABILITY = 481.wrf=default=default: CPORTABILITY = -DSPEC_CPU_LINUX -DSPEC_CPU_CASE_FLAG FPORTABILITY = -fno-second-underscore 450.soplex=peak=default: PORTABILITY = ################################################################# # # # SPECfp Tuning # # # ################################################################# fp=base: COPTIMIZE = -Ofast -HP:bdt=2m:heap=2m CXXOPTIMIZE = -Ofast -static -INLINE:aggressive=on -HP:bdt=2m:heap=2m FOPTIMIZE = -Ofast -apo -LNO:parallel_overhead=10000:fusion_peeling_limit=0 -HP:bdt=2m:heap=2m sw_base_ptrsize = 64-bit sw_peak_ptrsize = 32/64-bit ##################################################################### # FP Peak Tuning ##################################################################### fp=peak=default: COPTIMIZE = -Ofast -HP:bdt=2m:heap=2m CXXOPTIMIZE = -Ofast -INLINE:aggressive=on -HP:bdt=2m:heap=2m FOPTIMIZE = -Ofast -apo -LNO:parallel_overhead=10000:fusion_peeling_limit=0 -HP:bdt=2m:heap=2m PASS1_CFLAGS = -fb_create fbdata PASS2_CFLAGS = -fb_opt fbdata PASS1_CXXFLAGS = -fb_create fbdata PASS2_CXXFLAGS = -fb_opt fbdata PASS1_FFLAGS = -fb_create fbdata PASS2_FFLAGS = -fb_opt fbdata PASS1_LDFLAGS = -fb_create fbdata PASS2_LDFLAGS = -fb_opt fbdata 410.bwaves=peak=default: FOPTIMIZE = -Ofast -apo -OPT:malloc_alg=2 -CG:use_prefetchnta=on:cmp_peep=on -LNO:blocking=off:prefetch=3:prefetch_ahead=5:ignore_feedback=off:apo_use_feedback=on -WOPT:aggstr=0 feedback = 0 %if '%{cores_per_socket}' eq '6' % if '%{num_sockets}' eq '1' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3,4,5 bash run.sh % elif '%{num_sockets}' eq '2' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3,4,5,6,7,8,9,10,11 bash run.sh % endif #num_sockets %elif '%{cores_per_socket}' eq '4' % if '%{num_sockets}' eq '1' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3 bash run.sh % elif '%{num_sockets}' eq '2' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3,4,5,6,7 bash run.sh % endif #num_sockets %endif #cores_per_socket 416.gamess=peak=default: FOPTIMIZE = -O3 -LNO:fu=6:blocking=0:prefetch=0 -OPT:Ofast:ro=3:unroll_size=256 -HP:bdt=2m:heap=2m 433.milc=peak=default: COPTIMIZE = -Ofast -apo -CG:movnti=1:local_sched_alg=1 -CG:locs_shallow_depth=1:compute_to=on -HP:bdt=2m:heap=2m -LNO:prefetch=3 feedback = 0 434.zeusmp=peak=default: FOPTIMIZE = -Ofast -apo -LNO:blocking=off:interchange=off:fusion_peeling_limit=0 -OPT:treeheight=on:unroll_size=256 -CG:cmp_peep=on:compute_to=on -GRA:prioritize_by_density=on -HP:bdt=2m:heap=2m feedback = 0 435.gromacs=peak=default: FOPTIMIZE = -Ofast -apo -OPT:rsqrt=2 -HP:bdt=2m:heap=2m COPTIMIZE = $(FOPTIMIZE) feedback = 0 436.cactusADM=peak=default: FOPTIMIZE = -Ofast -apo -LANG:heap_allocation_threshold=1000 -LNO:prefetch_ahead=1 -HP:bdt=2m:heap=2m COPTIMIZE = $(FOPTIMIZE) %if '%{cores_per_socket}' eq '4' % if '%{num_sockets}' eq '1' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3 bash run.sh % elif '%{num_sockets}' eq '2' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3,4,5,6,7 bash run.sh % endif #num_sockets %endif #cores_per_socket 437.leslie3d=peak=default: FOPTIMIZE = -Ofast -apo -OPT:unroll_size=256 -LNO:prefetch_ahead=4:parallel_overhead=32768 -GRA:prioritize_by_density=on -m3dnow -HP:bdt=2m:heap=2m feedback = 0 444.namd=peak=default: CXXOPTIMIZE = -Ofast -LNO:ignore_feedback=off -CG:local_sched_alg=2:load_exe=0:compute_to=on -OPT:unroll_size=256 -fno-exceptions -HP:bdt=2m:heap=2m 447.dealII=peak=default: CXXOPTIMIZE = -Ofast -static -INLINE:aggressive=on -LNO:opt=0 -fno-emit-exceptions -m32 -OPT:unroll_times_max=8:unroll_size=256:unroll_level=2 -HP:bdt=2m:heap=2m -GRA:unspill=on -CG:cmp_peep=on -TENV:frame_pointer=off feedback = 0 450.soplex=peak=default: CXXOPTIMIZE = -O3 -INLINE:aggressive=on -OPT:IEEE_arith=3:IEEE_NaN_Inf=off:fold_unsigned_relops=on -CG:load_exe=0 -fno-exceptions -m32 -HP:bdt=2m:heap=2m 453.povray=peak=default: CXXOPTIMIZE = -Ofast -INLINE:aggressive=on -HP:bdt=2m:heap=2m 454.calculix=peak=default: FOPTIMIZE = -Ofast -apo -LNO:prefetch_ahead=30 -CG:load_exe=0:ptr_load_use=0:local_sched_alg=2:compute_to=on -WOPT:unroll=2 -GRA:optimize_boundary=on -HP:bdt=2m:heap=2m COPTIMIZE = -Ofast -LNO:prefetch_ahead=30 -CG:load_exe=0:ptr_load_use=0:local_sched_alg=2:compute_to=on -WOPT:unroll=2 -GRA:optimize_boundary=on -HP:bdt=2m:heap=2m feedback = 0 459.GemsFDTD=peak=default: FOPTIMIZE = -Ofast -apo -LNO:fission=2:prefetch_ahead=1 -CG:load_exe=0:local_sched_alg=1 -HP feedback = 0 465.tonto=peak=default: FOPTIMIZE = -Ofast -apo -OPT:alias=no_f90_pointer_alias -LNO:blocking=off -CG:load_exe=1 -IPA:plimit=525 -HP feedback = 0 470.lbm=peak=default: COPTIMIZE = -Ofast -mso -apo -CG:sse_cse_regs=0 -LNO:prefetch_ahead=4 -CG:locs_shallow_depth=1:cmp_peep=on:compute_to=on -OPT:unroll_times_max=8:unroll_size=256:unroll_level=2:keep_ext=on:alias=restricted -m3dnow -IPA:inline=off feedback = 0 %if '%{cores_per_socket}' eq '6' % if '%{num_sockets}' eq '1' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3,4,5 bash run.sh % elif '%{num_sockets}' eq '2' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3,4,5,6,7,8,9,10,11 bash run.sh % endif #num_sockets %elif '%{cores_per_socket}' eq '4' % if '%{num_sockets}' eq '1' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3 bash run.sh % elif '%{num_sockets}' eq '2' submit = echo "$command" > run.sh ; numactl -l --physcpubind=0,1,2,3,4,5,6,7 bash run.sh % endif #num_sockets %endif #cores_per_socket 481.wrf=peak=default: basepeak = 1 feedback = 0 482.sphinx3=peak=default: COPTIMIZE = -Ofast -OPT:malloc_alg=2 -CG:sse_cse_regs=0:locs_shallow_depth=1:cmp_peep=on:local_sched_alg=1 -INLINE:aggressive=on # 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: flagsurl001 = http://www.spec.org/cpu2006/flags/amd-platform-speed-revA.xml notes_000 =Environment variables set by runspec before the start of the run: notes_005 =LD_LIBRARY_PATH = "/root/work/cpu2006/amd1002-speed-libs-revA/64:/root/work/cpu2006/amd1002-speed-libs-revA/32" notes_010 =O64_OMP_AFFINITY_MAP = "0,1,2,3,4,5,6,7" notes_015 =O64_OMP_SPIN_USER_LOCK = "true" notes_020 =