| CPU2006 license: | 6 | Test date: | Oct-2010 |
|---|---|---|---|
| Test sponsor: | Oracle Corporation | 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 | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Seconds | Ratio | Seconds | Ratio | Seconds | Ratio | 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 | 30.0 | 453 | 32.8 | 414 | 32.8 | 414 | 30.0 | 453 | 32.8 | 414 | 32.8 | 414 |
| 416.gamess | 1570 | 12.5 | 1571 | 12.5 | 1571 | 12.5 | 1462 | 13.4 | 1462 | 13.4 | 1462 | 13.4 |
| 433.milc | 465 | 19.7 | 464 | 19.8 | 466 | 19.7 | 308 | 29.8 | 308 | 29.8 | 308 | 29.8 |
| 434.zeusmp | 279 | 32.6 | 279 | 32.6 | 279 | 32.6 | 266 | 34.2 | 266 | 34.2 | 266 | 34.1 |
| 435.gromacs | 1343 | 5.31 | 1342 | 5.32 | 1340 | 5.33 | 327 | 21.9 | 327 | 21.9 | 327 | 21.9 |
| 436.cactusADM | 44.3 | 270 | 44.9 | 266 | 43.7 | 273 | 40.2 | 297 | 40.5 | 295 | 40.8 | 293 |
| 437.leslie3d | 131 | 71.8 | 131 | 71.8 | 131 | 71.9 | 128 | 73.6 | 128 | 73.5 | 128 | 73.4 |
| 444.namd | 606 | 13.2 | 606 | 13.2 | 606 | 13.2 | 587 | 13.7 | 587 | 13.7 | 587 | 13.7 |
| 447.dealII | 516 | 22.2 | 516 | 22.2 | 517 | 22.1 | 481 | 23.8 | 481 | 23.8 | 481 | 23.8 |
| 450.soplex | 667 | 12.5 | 668 | 12.5 | 667 | 12.5 | 538 | 15.5 | 538 | 15.5 | 538 | 15.5 |
| 453.povray | 398 | 13.4 | 398 | 13.4 | 399 | 13.3 | 323 | 16.5 | 324 | 16.4 | 324 | 16.4 |
| 454.calculix | 453 | 18.2 | 452 | 18.2 | 453 | 18.2 | 393 | 21.0 | 394 | 21.0 | 394 | 21.0 |
| 459.GemsFDTD | 113 | 94.3 | 112 | 95.0 | 113 | 94.1 | 113 | 94.3 | 112 | 95.0 | 113 | 94.1 |
| 465.tonto | 729 | 13.5 | 731 | 13.5 | 729 | 13.5 | 453 | 21.7 | 453 | 21.7 | 454 | 21.7 |
| 470.lbm | 35.2 | 391 | 35.0 | 392 | 35.3 | 389 | 24.0 | 572 | 24.5 | 560 | 24.1 | 569 |
| 481.wrf | 636 | 17.5 | 637 | 17.5 | 636 | 17.6 | 452 | 24.7 | 452 | 24.7 | 451 | 24.7 |
| 482.sphinx3 | 762 | 25.6 | 762 | 25.6 | 763 | 25.6 | 719 | 27.1 | 719 | 27.1 | 719 | 27.1 |
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 = "64" SUNW_MP_PROCBIND = "127 125 123 121 119 117 115 113 111 109 107 105 103 101 99 97 95 93 91 89 87 85 83 81 79 77 75 73 71 69 67 65 63 61 59 57 55 53 51 49 47 45 43 41 39 37 35 33 31 29 27 25 23 21 19 17 15 13 11 9 7 5 3 1 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 90 92 94 96 98 100 102 104 106 108 110 112 114 116 118 120 122 124 126" 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 -xpagesize=4M -xipo=2 -xprefetch=latx:2 -xprefetch_level=3 -xprefetch_auto_type=indirect_array_access -xautopar -xreduction |
C++ benchmarks:
Fortran benchmarks:
| -fast -fma=fused -xpagesize=4M -xipo=2 -xprefetch=latx:2 -xprefetch_level=2 -xautopar -xreduction |
Benchmarks using both Fortran and C:
| -fast(cc) -fast(f90) -fma=fused -xpagesize=4M -xipo=2 -xprefetch=latx:2 -xprefetch_level=3 -xprefetch_auto_type=indirect_array_access -xautopar -xreduction -xprefetch_level=2 |
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:
C++ benchmarks:
Fortran benchmarks:
| 410.bwaves: | basepeak = yes |
| 416.gamess: | -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast -fma=fused -xpagesize=4M -xipo=2 -xprefetch_level=3 |
| 434.zeusmp: | -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast -fma=fused -xpagesize=4M -xipo=2 -xautopar -xreduction |
| 437.leslie3d: | Same as 434.zeusmp |
| 459.GemsFDTD: | basepeak = yes |
| 465.tonto: | -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast -xpagesize=4M -xipo=2 -fma=fused -xautopar -xreduction -Qoption iropt -Aparallel:nthreads=4 -lumem |
Benchmarks using both Fortran and C:
| 435.gromacs: | -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast(cc) -fast(f90) -fma=fused -xipo=2 -xpagesize=4M -xautopar -xreduction -Qoption iropt -Aparallel:nthreads=4 |
| 436.cactusADM: | -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast(cc) -fast(f90) -fma=fused -xpagesize=4M -xipo=2 -xautopar -xreduction |
| 454.calculix: | -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast(cc) -fast(f90) -fma=fused -xpagesize=4M -xipo=2 -xprefetch_level=3 -xprefetch=latx:3.0 -xalias_level=std -xautopar -xreduction -Qoption iropt -Aparallel:nthreads=4 |
| 481.wrf: | -fast(cc) -fast(f90) -fma=fused -xpagesize=4M -xipo=2 -xprefetch_level=3 -xunroll=8 -xautopar -xreduction -Qoption iropt -Aparallel:nthreads=12 |