| CPU2006 license: | 9006 | Test date: | Feb-2014 |
|---|---|---|---|
| Test sponsor: | NEC Corporation | Hardware Availability: | Mar-2014 |
| Tested by: | NEC Corporation | Software Availability: | Oct-2013 |
| Hardware | |
|---|---|
| CPU Name: | Intel Xeon E7-4890 v2 |
| CPU Characteristics: | Intel Turbo Boost Technology up to 3.40 GHz |
| CPU MHz: | 2800 |
| FPU: | Integrated |
| CPU(s) enabled: | 60 cores, 4 chips, 15 cores/chip, 2 threads/core |
| CPU(s) orderable: | 2,3,4 chips |
| Primary Cache: | 32 KB I + 32 KB D on chip per core |
| Secondary Cache: | 256 KB I+D on chip per core |
| L3 Cache: | 37.5 MB I+D on chip per chip |
| Other Cache: | None |
| Memory: | 1 TB (64 x 16 GB 2Rx4 PC3L-12800R-11, ECC, running at 1333 MHz and CL9) |
| Disk Subsystem: | 1 x 300 GB SAS, 10000 RPM |
| Other Hardware: | None |
| Software | |
|---|---|
| Operating System: | Red Hat Enterprise Linux Server release 6.4 (Santiago) Kernel 2.6.32-358.23.2.el6.x86_64 |
| Compiler: | C/C++: Version 14.0.1.106 of Intel C++ Studio XE for Linux |
| Auto Parallel: | No |
| File System: | ext4 |
| System State: | Run level 3 (multi-user) |
| Base Pointers: | 32-bit |
| Peak Pointers: | 32/64-bit |
| Other Software: | Microquill SmartHeap Multi-Core V10.01 |
Results Table
| Benchmark | Base | Peak | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Copies | Seconds | Ratio | Seconds | Ratio | Seconds | Ratio | Copies | Seconds | Ratio | Seconds | Ratio | Seconds | Ratio | |
| Results appear in the order in which they were run. Bold underlined text indicates a median measurement. | ||||||||||||||
| 400.perlbench | 120 | 644 | 1820 | 645 | 1820 | 644 | 1820 | 120 | 538 | 2180 | 547 | 2140 | 543 | 2160 |
| 401.bzip2 | 120 | 990 | 1170 | 991 | 1170 | 990 | 1170 | 120 | 981 | 1180 | 981 | 1180 | 980 | 1180 |
| 403.gcc | 120 | 585 | 1650 | 586 | 1650 | 585 | 1650 | 120 | 586 | 1650 | 587 | 1650 | 587 | 1650 |
| 429.mcf | 120 | 383 | 2860 | 383 | 2860 | 382 | 2860 | 120 | 383 | 2860 | 383 | 2860 | 382 | 2860 |
| 445.gobmk | 120 | 697 | 1810 | 698 | 1800 | 699 | 1800 | 120 | 684 | 1840 | 683 | 1840 | 670 | 1880 |
| 456.hmmer | 120 | 346 | 3230 | 347 | 3230 | 345 | 3250 | 120 | 323 | 3460 | 323 | 3470 | 323 | 3460 |
| 458.sjeng | 120 | 823 | 1760 | 823 | 1770 | 824 | 1760 | 120 | 786 | 1850 | 785 | 1850 | 784 | 1850 |
| 462.libquantum | 120 | 163 | 15300 | 162 | 15300 | 162 | 15300 | 120 | 163 | 15300 | 162 | 15300 | 162 | 15300 |
| 464.h264ref | 120 | 867 | 3060 | 849 | 3130 | 868 | 3060 | 120 | 849 | 3130 | 859 | 3090 | 859 | 3090 |
| 471.omnetpp | 120 | 765 | 981 | 765 | 980 | 764 | 981 | 120 | 733 | 1020 | 734 | 1020 | 733 | 1020 |
| 473.astar | 120 | 674 | 1250 | 673 | 1250 | 675 | 1250 | 120 | 674 | 1250 | 673 | 1250 | 675 | 1250 |
| 483.xalancbmk | 120 | 364 | 2270 | 363 | 2280 | 362 | 2280 | 120 | 364 | 2270 | 363 | 2280 | 362 | 2280 |
Submit Notes
The numactl mechanism was used to bind copies to processors. The config file option 'submit' was used to generate numactl commands to bind each copy to a specific processor. For details, please see the config file.
Operating System Notes
Stack size set to unlimited using "ulimit -s unlimited" Zone reclaim mode enabled with: echo 1 > /proc/sys/vm/zone_reclaim_mode
Platform Notes
BIOS Settings: Memory RAS Mode: Independent mode
General Notes
Environment variables set by runspec before the start of the run: KMP_AFFINITY = "granularity=fine,scatter" KMP_STACKSIZE = "200M" LD_LIBRARY_PATH = "/opt/SmartHeap_10mc/lib:/opt/SmartHeap_10mc/lib64:/opt/intel/composer_xe_2013_sp1.1.106/compiler/lib/ia32:/opt/intel/composer_xe_2013_sp1.1.106/compiler/lib/intel64" OMP_NUM_THREADS = "120" Transparent Huge Pages enabled with: echo always > /sys/kernel/mm/redhat_transparent_hugepage/enabled Filesystem page cache cleared with: echo 1 > /proc/sys/vm/drop_caches runspec command invoked through numactl i.e.: numactl --interleave=all runspec <etc>
Base Portability Flags
| 400.perlbench: | -DSPEC_CPU_LINUX_IA32 |
| 462.libquantum: | -DSPEC_CPU_LINUX |
| 483.xalancbmk: | -DSPEC_CPU_LINUX |
Base Optimization Flags
C benchmarks:
| -xSSE4.2 -ipo -O3 -no-prec-div -opt-prefetch -opt-mem-layout-trans=3 |
C++ benchmarks:
| -xSSE4.2 -ipo -O3 -no-prec-div -opt-prefetch -opt-mem-layout-trans=3 -Wl,-z,muldefs -L/opt/SmartHeap_10mc/lib -lsmartheap |
Peak Compiler Invocation
C benchmarks (except as noted below):
| icc -m32 | |
| 400.perlbench: | icc -m64 |
| 401.bzip2: | icc -m64 |
| 456.hmmer: | icc -m64 |
| 458.sjeng: | icc -m64 |
C++ benchmarks:
| icpc -m32 |
Peak Portability Flags
| 400.perlbench: | -DSPEC_CPU_LP64 -DSPEC_CPU_LINUX_X64 |
| 401.bzip2: | -DSPEC_CPU_LP64 |
| 456.hmmer: | -DSPEC_CPU_LP64 |
| 458.sjeng: | -DSPEC_CPU_LP64 |
| 462.libquantum: | -DSPEC_CPU_LINUX |
| 483.xalancbmk: | -DSPEC_CPU_LINUX |
Peak Optimization Flags
C benchmarks:
| 400.perlbench: | -xSSE4.2(pass 2) -prof-gen(pass 1) -ipo(pass 2) -O3(pass 2) -no-prec-div(pass 2) -prof-use(pass 2) -auto-ilp32 |
| 401.bzip2: | -xSSE4.2(pass 2) -prof-gen(pass 1) -ipo(pass 2) -O3(pass 2) -no-prec-div(pass 2) -prof-use(pass 2) -opt-prefetch -auto-ilp32 -ansi-alias |
| 403.gcc: | -xSSE4.2 -ipo -O3 -no-prec-div |
| 429.mcf: | basepeak = yes |
| 445.gobmk: | -xSSE4.2(pass 2) -prof-gen(pass 1) -prof-use(pass 2) -ansi-alias -opt-mem-layout-trans=3 |
| 456.hmmer: | -xSSE4.2 -ipo -O3 -no-prec-div -unroll2 -auto-ilp32 |
| 458.sjeng: | -xSSE4.2(pass 2) -prof-gen(pass 1) -ipo(pass 2) -O3(pass 2) -no-prec-div(pass 2) -prof-use(pass 2) -unroll4 -auto-ilp32 |
| 462.libquantum: | basepeak = yes |
| 464.h264ref: | -xSSE4.2(pass 2) -prof-gen(pass 1) -ipo(pass 2) -O3(pass 2) -no-prec-div(pass 2) -prof-use(pass 2) -unroll2 -ansi-alias |
C++ benchmarks:
| 471.omnetpp: | -xSSE4.2(pass 2) -prof-gen(pass 1) -ipo(pass 2) -O3(pass 2) -no-prec-div(pass 2) -prof-use(pass 2) -ansi-alias -opt-ra-region-strategy=block -Wl,-z,muldefs -L/opt/SmartHeap_10mc/lib -lsmartheap |
| 473.astar: | basepeak = yes |
| 483.xalancbmk: | basepeak = yes |