| CPU2006 license: | 11 | Test date: | Sep-2008 |
|---|---|---|---|
| Test sponsor: | IBM Corporation | Hardware Availability: | Nov-2008 |
| Tested by: | IBM Corporation | Software Availability: | Nov-2008 |
| Hardware | |
|---|---|
| CPU Name: | POWER6+ |
| CPU Characteristics: | |
| CPU MHz: | 5000 |
| FPU: | Integrated |
| CPU(s) enabled: | 16 cores, 8 chips, 2 cores/chip, 2 threads/core |
| CPU(s) orderable: | 2,4,8,12,16 cores |
| Primary Cache: | 64 KB I + 64 KB D on chip per core |
| Secondary Cache: | 4 MB I+D on chip per core |
| L3 Cache: | 32 MB I+D off chip per chip |
| Other Cache: | None |
| Memory: | 128 GB (64x2 GB) DDR2 667 MHz |
| Disk Subsystem: | 4x73 GB 4x146 GB SAS 15K RPM |
| Other Hardware: | None |
| Software | |
|---|---|
| Operating System: | IBM AIX V6.1 with the 6100-02 Technology Level |
| Compiler: | IBM XL C/C++ V10.1 for AIX IBM XL Fortran V12.1 for AIX |
| Auto Parallel: | No |
| File System: | AIX/JFS2 |
| System State: | Multi-user |
| Base Pointers: | 32-bit |
| Peak Pointers: | 32/64-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 | 32 | 553 | 786 | 548 | 793 | 550 | 790 | 32 | 526 | 827 | 526 | 827 | 526 | 827 |
| 416.gamess | 32 | 1568 | 400 | 1569 | 399 | 1569 | 399 | 32 | 1528 | 410 | 1528 | 410 | 1529 | 410 |
| 433.milc | 32 | 672 | 437 | 673 | 437 | 672 | 437 | 32 | 655 | 448 | 655 | 449 | 655 | 448 |
| 434.zeusmp | 32 | 840 | 347 | 843 | 345 | 840 | 347 | 32 | 514 | 566 | 496 | 587 | 496 | 588 |
| 435.gromacs | 32 | 846 | 270 | 847 | 270 | 847 | 270 | 32 | 550 | 415 | 539 | 424 | 551 | 415 |
| 436.cactusADM | 32 | 991 | 386 | 990 | 386 | 990 | 386 | 32 | 670 | 571 | 638 | 600 | 636 | 601 |
| 437.leslie3d | 32 | 590 | 510 | 610 | 493 | 609 | 494 | 16 | 304 | 495 | 309 | 486 | 304 | 495 |
| 444.namd | 32 | 495 | 519 | 496 | 517 | 498 | 515 | 32 | 491 | 523 | 495 | 519 | 491 | 522 |
| 447.dealII | 32 | 430 | 852 | 428 | 854 | 431 | 850 | 32 | 430 | 852 | 428 | 854 | 431 | 850 |
| 450.soplex | 32 | 602 | 444 | 545 | 489 | 549 | 486 | 32 | 602 | 444 | 545 | 489 | 549 | 486 |
| 453.povray | 32 | 515 | 330 | 516 | 330 | 515 | 330 | 32 | 372 | 457 | 373 | 456 | 373 | 456 |
| 454.calculix | 32 | 628 | 420 | 628 | 420 | 629 | 420 | 32 | 566 | 467 | 564 | 468 | 561 | 470 |
| 459.GemsFDTD | 32 | 917 | 370 | 918 | 370 | 918 | 370 | 32 | 903 | 376 | 883 | 384 | 901 | 377 |
| 465.tonto | 32 | 1056 | 298 | 1055 | 299 | 1055 | 299 | 32 | 586 | 538 | 585 | 538 | 585 | 538 |
| 470.lbm | 32 | 547 | 803 | 547 | 804 | 547 | 804 | 32 | 546 | 805 | 547 | 804 | 546 | 805 |
| 481.wrf | 32 | 729 | 490 | 730 | 490 | 736 | 485 | 32 | 672 | 532 | 670 | 533 | 667 | 536 |
| 482.sphinx3 | 32 | 891 | 700 | 877 | 711 | 869 | 718 | 32 | 849 | 735 | 841 | 741 | 852 | 732 |
Peak Tuning Notes
fdpr binary optimization tool used for 410.bwaves
with options -bf -bp -ece -lap -las -nop -nopr -pto -RC -RD -rmte -so -tlo -A 64
-lu 6 -rt 0.10 -ihf 60 -sdpla 32 -sdpms 32 -shci 10 -si -siht 15 -lun 32
fdpr binary optimization tool used for 433.milc 435.gromacs 437.leslie3d 453.povray
454.calculix 481.wrf 482.sphinx3
with options -O4 -vrox -pbsi
fdpr binary optimization tool used for 434.zeusmp 470.lbm
with options -O3 -vrox -sdp 9
fdpr binary optimization tool used for 459.GemsFDTD
with options -bf -bp -ece -hr -lap -nop -pca -RC -rmte -si -tb -tlo -vro -A 32 -rt 0.80
-hrf 0.05 -sdp 5 -sdpms 512 -shci 90 -lun 27 -rcctf 0.70 -rccrf 0.80 -rcaf 2
Submit Notes
The config file option 'submit' was used to assign benchmark copy to specific kernel thread using the "bindprocessor" command (see flags file for details).
Operating System Notes
all ulimits set to unlimited. 3200 16M large pages defined with vmo command
Platform Notes
System set to "Enhanced" mode when defining partition on HMC.
General Notes
Environment variables set by runspec before the start of the run: MALLOCOPTIONS = "pool" MEMORY_AFFINITY = "MCM" XLFRTEOPTS = "intrinthds=1" See the flags file for details on settings.
Base Compiler Invocation
C benchmarks:
| /usr/vac/bin/xlc -qlanglvl=extc99 |
C++ benchmarks:
| /usr/vacpp/bin/xlC |
Fortran benchmarks:
| /usr/bin/xlf95 |
Benchmarks using both Fortran and C:
| /usr/vac/bin/xlc -qlanglvl=extc99 /usr/bin/xlf95 |
Base Portability Flags
| 410.bwaves: | -qfixed |
| 416.gamess: | -qfixed |
| 434.zeusmp: | -qfixed |
| 435.gromacs: | -qfixed -qextname |
| 436.cactusADM: | -qfixed -qextname |
| 437.leslie3d: | -qfixed |
| 454.calculix: | -qfixed -qextname |
| 481.wrf: | -DSPEC_CPU_AIX -DNOUNDERSCORE |
| 482.sphinx3: | -qchars=signed |
Base Optimization Flags
C benchmarks:
| -bmaxdata:0x40000000 -O5 -qlargepage -D_ILS_MACROS -blpdata |
C++ benchmarks:
| -bmaxdata:0x50000000 -O5 -qlargepage -D_ILS_MACROS -qrtti=all -D__IBM_FAST_VECTOR -D__IBM_FAST_SET_MAP_ITERATOR -blpdata |
Fortran benchmarks:
| -bmaxdata:0x60000000 -O5 -qlargepage -qsmallstack=dynlenonheap -qalias=nostd -blpdata |
Benchmarks using both Fortran and C:
| -bmaxdata:0x60000000 -O5 -qlargepage -D_ILS_MACROS -qsmallstack=dynlenonheap -qalias=nostd -blpdata |
Base Other Flags
C benchmarks:
| -qipa=threads -qipa=noobject -qsuppress=1500-036 |
C++ benchmarks:
| -qipa=threads -qipa=noobject -qsuppress=1500-036 |
Fortran benchmarks:
| -qipa=threads -qipa=noobject -qsuppress=1500-010 -qsuppress=cmpmsg -qsuppress=1500-036 |
Benchmarks using both Fortran and C:
| -qipa=threads -qipa=noobject -qsuppress=1500-010 -qsuppress=cmpmsg -qsuppress=1500-036 |
Peak Compiler Invocation
C benchmarks:
| /usr/vac/bin/xlc -qlanglvl=extc99 |
C++ benchmarks:
| /usr/vacpp/bin/xlC |
Fortran benchmarks:
| /usr/bin/xlf95 |
Benchmarks using both Fortran and C:
| /usr/vac/bin/xlc -qlanglvl=extc99 /usr/bin/xlf95 |
Peak Portability Flags
| 410.bwaves: | -qfixed |
| 416.gamess: | -qfixed |
| 434.zeusmp: | -qfixed |
| 435.gromacs: | -qfixed -qextname |
| 436.cactusADM: | -qfixed -qextname |
| 437.leslie3d: | -qfixed |
| 454.calculix: | -qfixed -qextname |
| 481.wrf: | -DSPEC_CPU_AIX -DNOUNDERSCORE |
| 482.sphinx3: | -qchars=signed |
Peak Optimization Flags
C benchmarks:
| 433.milc: | -bmaxdata:0x40000000 -O5 -qlargepage -D_ILS_MACROS -qalign=natural -qfdpr -blpdata |
| 470.lbm: | -qpdf1(pass 1) -qpdf2(pass 2) -O3 -qarch=auto -qtune=auto -qlargepage -q64 -D_ILS_MACROS -qfdpr -blpdata |
| 482.sphinx3: | -qpdf1(pass 1) -qpdf2(pass 2) -O4 -qlargepage -D_ILS_MACROS -qfdpr -blpdata |
C++ benchmarks:
| 444.namd: | -qpdf1(pass 1) -qpdf2(pass 2) -O5 -qlargepage -D_ILS_MACROS -blpdata |
| 447.dealII: | basepeak = yes |
| 450.soplex: | basepeak = yes |
| 453.povray: | -qpdf1(pass 1) -qpdf2(pass 2) -O5 -D_ILS_MACROS -qalign=natural -qfdpr |
Fortran benchmarks:
| 410.bwaves: | -bmaxdata:0x50000000 -O5 -qlargepage -qenablevmx -qvecnvol -qfdpr -qsmallstack=dynlenonheap -blpdata |
| 416.gamess: | -bmaxdata:0x40000000 -qpdf1(pass 1) -qpdf2(pass 2) -O5 -qlargepage -qalias=nostd -blpdata |
| 434.zeusmp: | -bmaxdata:0x40000000 -qpdf1(pass 1) -qpdf2(pass 2) -O3 -qarch=auto -qtune=auto -qlargepage -qenablevmx -qvecnvol -qxlf90=nosignedzero -qfdpr -blpdata |
| 437.leslie3d: | -O5 -qlargepage -qenablevmx -qvecnvol -qfdpr -blpdata |
| 459.GemsFDTD: | -O4 -qlargepage -q64 -qfdpr -blpdata |
| 465.tonto: | -bmaxdata:0x50000000 -qpdf1(pass 1) -qpdf2(pass 2) -O5 -bdatapsize:64K -bstackpsize:64K -btextpsize:64K |
Benchmarks using both Fortran and C:
| 435.gromacs: | -qpdf1(pass 1) -qpdf2(pass 2) -O5 -D_ILS_MACROS -qfdpr |
| 436.cactusADM: | -bmaxdata:0x60000000 -qpdf1(pass 1) -qpdf2(pass 2) -O2 -qarch=auto -qtune=auto -qenablevmx -qvecnvol -D_ILS_MACROS -qfdpr -qnostrict -bdatapsize:64K -bstackpsize:64K -btextpsize:64K |
| 454.calculix: | -O4 -qlargepage -q64 -D_ILS_MACROS -qfdpr -blpdata |
| 481.wrf: | -qpdf1(pass 1) -qpdf2(pass 2) -O5 -qlargepage -q64 -D_ILS_MACROS -qfdpr -blpdata |
Peak Other Flags
C benchmarks:
| -qipa=threads -qipa=noobject -qsuppress=1500-036 |
C++ benchmarks:
| -qipa=threads -qipa=noobject -qsuppress=1500-036 |
Fortran benchmarks:
| -qipa=threads -qipa=noobject -qsuppress=1500-010 -qsuppress=cmpmsg -qsuppress=1500-036 |
Benchmarks using both Fortran and C:
| -qipa=threads -qipa=noobject -qsuppress=1500-010 -qsuppress=cmpmsg -qsuppress=1500-036 |