| CPU2006 license: | 19 | Test date: | Nov-2010 |
|---|---|---|---|
| Test sponsor: | Fujitsu | Hardware Availability: | Dec-2010 |
| Tested by: | Oracle Corporation | Software Availability: | Sep-2010 |
| Hardware | |
|---|---|
| CPU Name: | SPARC64 VII+ |
| CPU Characteristics: | |
| CPU MHz: | 3000 |
| FPU: | Integrated |
| CPU(s) enabled: | 64 cores, 16 chips, 4 cores/chip, 2 threads/core |
| CPU(s) orderable: | 1 to 4 CMUs; each CMU contains 2 or 4 CPU chips |
| Primary Cache: | 64 KB I + 64 KB D on chip per core |
| Secondary Cache: | 12 MB I+D on chip per chip |
| L3 Cache: | None |
| Other Cache: | None |
| Memory: | 512 GB (128 x 4 GB, 8-way interleaved) |
| Disk Subsystem: | 698 GB mirrored partition on 12 x 146 GB 15K RPM SAS disks in each of 2 StorageTek 2530 Array (24 total disk, 12 in each array) |
| Other Hardware: | None |
| Software | |
|---|---|
| Operating System: | Oracle Solaris 10 9/10 |
| Compiler: | Oracle Solaris Studio 12.2 |
| Auto Parallel: | Yes |
| File System: | ufs |
| System State: | Default |
| Base Pointers: | 32-bit |
| Peak Pointers: | 32-bit |
| Other Software: | None |
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. | ||||||||||||||
| 410.bwaves | 127 | 3235 | 533 | 3191 | 541 | 3192 | 541 | 127 | 3191 | 541 | 3192 | 541 | 3191 | 541 |
| 416.gamess | 127 | 2645 | 940 | 2632 | 945 | 2705 | 919 | 127 | 2499 | 995 | 2498 | 996 | 2527 | 984 |
| 433.milc | 127 | 2297 | 508 | 2298 | 507 | 2297 | 507 | 127 | 2297 | 508 | 2298 | 507 | 2297 | 507 |
| 434.zeusmp | 127 | 1441 | 802 | 1440 | 802 | 1444 | 801 | 127 | 1441 | 802 | 1431 | 808 | 1430 | 808 |
| 435.gromacs | 127 | 900 | 1010 | 888 | 1020 | 893 | 1020 | 127 | 840 | 1080 | 831 | 1090 | 829 | 1090 |
| 436.cactusADM | 127 | 1426 | 1060 | 1425 | 1070 | 1430 | 1060 | 63 | 633 | 1190 | 632 | 1190 | 628 | 1200 |
| 437.leslie3d | 127 | 2881 | 414 | 2880 | 415 | 2879 | 415 | 63 | 1399 | 423 | 1399 | 423 | 1399 | 423 |
| 444.namd | 127 | 981 | 1040 | 942 | 1080 | 975 | 1050 | 127 | 844 | 1210 | 844 | 1210 | 844 | 1210 |
| 447.dealII | 127 | 912 | 1590 | 911 | 1590 | 912 | 1590 | 127 | 896 | 1620 | 890 | 1630 | 910 | 1600 |
| 450.soplex | 127 | 2958 | 358 | 2954 | 359 | 2960 | 358 | 64 | 1392 | 384 | 1377 | 388 | 1375 | 388 |
| 453.povray | 127 | 596 | 1130 | 600 | 1130 | 597 | 1130 | 127 | 429 | 1580 | 429 | 1580 | 428 | 1580 |
| 454.calculix | 127 | 769 | 1360 | 765 | 1370 | 772 | 1360 | 127 | 704 | 1490 | 703 | 1490 | 704 | 1490 |
| 459.GemsFDTD | 127 | 4141 | 325 | 4153 | 324 | 4141 | 325 | 127 | 4141 | 325 | 4153 | 324 | 4141 | 325 |
| 465.tonto | 127 | 1214 | 1030 | 1204 | 1040 | 1218 | 1030 | 127 | 1158 | 1080 | 1160 | 1080 | 1170 | 1070 |
| 470.lbm | 127 | 5210 | 335 | 5210 | 335 | 5210 | 335 | 1 | 21.5 | 640 | 21.5 | 638 | 21.5 | 639 |
| 481.wrf | 127 | 2174 | 653 | 2177 | 652 | 2169 | 654 | 127 | 2131 | 666 | 2128 | 667 | 2131 | 666 |
| 482.sphinx3 | 127 | 5049 | 490 | 5049 | 490 | 5042 | 491 | 64 | 2415 | 517 | 2423 | 515 | 2418 | 516 |
Compiler Invocation Notes
Oracle Solaris Studio 12.2 is distributed with mandatory OS patches 118683-05 119963-20 120753-08 Oracle Solaris Studio 12.2 and patches are available at http://oracle.com/goto/solarisstudio The Apache C++ Standard Library V4.2.1 was installed from http://stdcxx.apache.org/download.html using: alias gmake=specmake gmake BUILDTYPE=8d CONFIG=sunpro.config
Submit Notes
Processes were assigned to specific processors using 'pbind' commands. The config file option 'submit' was used, along with a list of processors in the 'BIND' variable, to generate the pbind commands. (For details, please see the config file.)
Operating System Notes
ulimit -s 131072 was used to limit the space
consumed by the stack (and therefore make more
space available to the heap).
/etc/system parameters
autoup=600
Causes pages older than the listed number of seconds to
be written by fsflush.
tune_t_fsflushr=10
Controls how many seconds elapse between runs of the
page flush daemon, fsflush.
lpg_alloc_prefer=1
Indicates that extra effort should be taken to ensure
that pages are created in the nearby lgroup (NUMA location).
The "webconsole" service was turned off using
svcadm disable webconsole
The system had 75 GB of swap space.
Platform Notes
Memory is 8-way interleaved by filling all slots with the same capacity DIMMs. This result is measured on a SPARC Enterprise M8000 server from Oracle. The SPARC Enterprise M8000 server from Oracle and from Fujitsu are electrically equivalent.
General Notes
Environment variables set by runspec before the start of the run: OMP_NUM_THREADS = "128" SUNW_MP_PROCBIND = "127 126 125 124 123 122 121 120 119 118 117 116 115 114 113 112 111 110 109 108 107 106 105 104 103 102 101 100 99 98 97 96 95 94 93 92 91 90 89 88 87 86 85 84 83 82 81 80 79 78 77 76 75 74 73 72 71 70 69 68 67 66 65 64 63 62 61 60 59 58 57 56 55 54 53 52 51 50 49 48 47 46 45 44 43 42 41 40 39 38 37 36 35 34 33 32 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0" SUNW_MP_THR_IDLE = "SPIN" 447.dealII (peak): "apache_stdcxx_4_2_1" src.alt was used. 447.dealII (base): "apache_stdcxx_4_2_1" src.alt was used.
Base Compiler Invocation
C benchmarks:
| cc |
C++ benchmarks:
| CC |
Fortran benchmarks:
| f90 |
Benchmarks using both Fortran and C:
| cc f90 |
Base Optimization Flags
C benchmarks:
| -fast -fma=fused -xipo=2 -xpagesize=4M -xalias_level=std -xprefetch_auto_type=indirect_array_access -xprefetch_level=3 -fsimple=1 |
C++ benchmarks:
Fortran benchmarks:
| -fast -fma=fused -xipo=2 -xpagesize=4M |
Benchmarks using both Fortran and C:
| -fast(cc) -fast(f90) -fma=fused -xipo=2 -xpagesize=4M -xalias_level=std -xprefetch_auto_type=indirect_array_access -xprefetch_level=3 -fsimple=1 |
Base Other Flags
C benchmarks:
| -xjobs=32 -V -# |
C++ benchmarks:
| -xjobs=32 -verbose=diags,version |
Fortran benchmarks:
| -xjobs=32 -V -v |
Benchmarks using both Fortran and C:
| -xjobs=32 -V -# -v |
Peak Compiler Invocation
C benchmarks:
| cc |
C++ benchmarks:
| CC |
Fortran benchmarks:
| f90 |
Benchmarks using both Fortran and C:
| cc f90 |
Peak Optimization Flags
C benchmarks:
| 433.milc: | basepeak = yes |
| 470.lbm: | -fast -xpagesize=4M -fma=fused -xipo=2 -xprefetch_level=3 -xvector -xarch=generic -xautopar -xreduction |
| 482.sphinx3: | -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast -xpagesize=4M -fma=fused -xipo=2 -xinline= -xalias_level=strong -xprefetch_level=2 -lfast -ll2amm |
C++ benchmarks:
Fortran benchmarks:
| 410.bwaves: | -fast -xpagesize=4M -fma=fused -xipo=2 |
| 416.gamess: | -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast -xpagesize=4M -fma=fused -xipo=2 -xprefetch=no%auto -xO3 |
| 434.zeusmp: | -fast -xpagesize=4M -fma=fused -xipo=2 -xprefetch_level=1 -ll2amm -xunroll=5 |
| 437.leslie3d: | -fast -xpagesize=4M -M /usr/lib/ld/map.bssalign -xprefetch=no |
| 459.GemsFDTD: | basepeak = yes |
| 465.tonto: | -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast -xpagesize=4M -xipo=2 -xprefetch=no -lfast -ll2amm |
Benchmarks using both Fortran and C:
| 435.gromacs: | -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast(cc) -fast(f90) -xpagesize=4M -fma=fused -xtarget=generic -xinline= -fsimple=0 -xlinkopt -xvector -xdepend |
| 436.cactusADM: | -fast(cc) -fast(f90) -xpagesize=4M -fma=fused -xipo=2 |
| 454.calculix: | -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast(cc) -fast(f90) -xpagesize=4M -fma=fused -xipo=2 -xvector -xprefetch=latx:3 -xalias_level=std |
| 481.wrf: | -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast(cc) -fast(f90) -xpagesize=4M -fma=fused -xipo=2 -xprefetch_level=3 -xprefetch_auto_type=indirect_array_access -ll2amm |