CPU2006 license: | 19 | Test date: | Apr-2007 |
---|---|---|---|
Test sponsor: | Fujitsu Limited | Hardware Availability: | Apr-2007 |
Tested by: | Sun Microsystems | Software Availability: | Jul-2007 |
Hardware | |
---|---|
CPU Name: | SPARC64 VI |
CPU Characteristics: | |
CPU MHz: | 2400 |
FPU: | Integrated |
CPU(s) enabled: | 128 cores, 64 chips, 2 cores/chip, 2 threads/core |
CPU(s) orderable: | 1 to 16 CMUs; each CMU contains 2 or 4 chips |
Primary Cache: | 128 KB I + 128 KB D on chip per core |
Secondary Cache: | 6 MB I+D on chip per chip |
L3 Cache: | None |
Other Cache: | None |
Memory: | 1 TB (512 x 2 GB) |
Disk Subsystem: | 792 GB RAID 1+0 created by Solaris Volume Manager with 24 x 73 GB 10,000 RPM Fujitsu MAY2073RC SAS |
Other Hardware: | None |
Software | |
---|---|
Operating System: | Solaris 10 7/07 (build s10s_u4wos_03) |
Compiler: | Sun Studio 12 (build 44.0) |
Auto Parallel: | No |
File System: | ufs |
System State: | Default |
Base Pointers: | 32-bit |
Peak Pointers: | 32-bit |
Other Software: | None |
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 | 255 | 2675 | 1300 | 2665 | 1300 | 2666 | 1300 | 255 | 2675 | 1300 | 2665 | 1300 | 2666 | 1300 |
416.gamess | 255 | 3862 | 1290 | 3861 | 1290 | 3863 | 1290 | 127 | 1836 | 1350 | 1834 | 1360 | 1832 | 1360 |
433.milc | 255 | 3769 | 621 | 3775 | 620 | 3774 | 620 | 255 | 3726 | 628 | 3731 | 627 | 3729 | 628 |
434.zeusmp | 255 | 2457 | 945 | 2456 | 945 | 2456 | 945 | 255 | 2457 | 945 | 2456 | 945 | 2456 | 945 |
435.gromacs | 255 | 1376 | 1320 | 1382 | 1320 | 1376 | 1320 | 127 | 635 | 1430 | 636 | 1430 | 636 | 1430 |
436.cactusADM | 255 | 1689 | 1800 | 1694 | 1800 | 1688 | 1810 | 255 | 1689 | 1800 | 1694 | 1800 | 1688 | 1810 |
437.leslie3d | 255 | 2157 | 1110 | 2160 | 1110 | 2158 | 1110 | 255 | 2099 | 1140 | 2095 | 1140 | 2106 | 1140 |
444.namd | 255 | 1558 | 1310 | 1543 | 1330 | 1536 | 1330 | 127 | 723 | 1410 | 721 | 1410 | 721 | 1410 |
447.dealII | 255 | 1635 | 1780 | 1637 | 1780 | 1646 | 1770 | 255 | 1580 | 1850 | 1575 | 1850 | 1575 | 1850 |
450.soplex | 255 | 2921 | 728 | 2804 | 759 | 2794 | 761 | 255 | 2772 | 767 | 2697 | 788 | 2684 | 792 |
453.povray | 255 | 1206 | 1120 | 1204 | 1130 | 1215 | 1120 | 127 | 378 | 1790 | 378 | 1790 | 377 | 1790 |
454.calculix | 255 | 1350 | 1560 | 1315 | 1600 | 1314 | 1600 | 255 | 1350 | 1560 | 1315 | 1600 | 1314 | 1600 |
459.GemsFDTD | 255 | 3443 | 786 | 3440 | 787 | 3440 | 786 | 255 | 3438 | 787 | 3428 | 789 | 3442 | 786 |
465.tonto | 255 | 1756 | 1430 | 1759 | 1430 | 1756 | 1430 | 127 | 802 | 1560 | 802 | 1560 | 800 | 1560 |
470.lbm | 255 | 3488 | 1000 | 3656 | 958 | 3658 | 958 | 255 | 3359 | 1040 | 3363 | 1040 | 3363 | 1040 |
481.wrf | 255 | 2017 | 1410 | 1948 | 1460 | 1991 | 1430 | 255 | 2017 | 1410 | 1948 | 1460 | 1991 | 1430 |
482.sphinx3 | 255 | 5199 | 956 | 5204 | 955 | 5202 | 955 | 255 | 5199 | 956 | 5204 | 955 | 5202 | 955 |
Processes were bound to cores using "submit" and "pbind". The SPEC toolset was bound to processor 0. These shell commands request use of local 4MB pages: export LD_PRELOAD=madv.so.1:mpss.so.1 export MPSSHEAP=4MB export MPSSSTACK=4MB export MADV=access_lwp 'access_lwp' means that the next light weight process to touch the specified address range will access it the most heavily. 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=300 Causes pages older than the listed number of seconds to be written by fsflush. bufhwm=3000 Memory byte limit for caching I/O buffers segmap_percent=1 Set maximum percent memory for file system cache tune_t_fsflushr=3 Controls how many seconds elapse between runs of the page flush daemon, fsflush. The "webconsole" service was turned off using svcadm disable webconsole
"CMU" = CPU/Memory Unit; each holds 2 or 4 CPU chips. Memory was 8-way interleaved by filling all slots with the same capacity DIMMs. This result was measured using a Sun SPARC Enterprise M9000 Server. Note that the Fujitsu SPARC Enterprise M9000 and Sun SPARC Enterprise M9000 are electrically equivalent.
cc |
CC |
f90 |
cc f90 |
-fast -fma=fused -xcache=128/64/2:6144/256/12 -xipo=2 -xpagesize=4M -xprefetch_level=2 -xprefetch=latx:2 -xalias_level=std -xprefetch_level=3 -xprefetch_auto_type=indirect_array_access |
-xdepend -library=stlport4 -fast -fma=fused -xcache=128/64/2:6144/256/12 -xipo=2 -xpagesize=4M -xprefetch_level=2 -xprefetch=latx:2 -xalias_level=compatible |
-fast -fma=fused -xcache=128/64/2:6144/256/12 -xipo=2 -xpagesize=4M -xprefetch_level=2 -xprefetch=latx:2 |
-fast(cc) -fast(f90) -fma=fused -xcache=128/64/2:6144/256/12 -xipo=2 -xpagesize=4M -xprefetch_level=2 -xprefetch=latx:2 -xalias_level=std -xprefetch_level=3 -xprefetch_auto_type=indirect_array_access |
-xjobs=24 -V -# |
-xjobs=24 -verbose=diags,version |
-xjobs=24 -V -v |
-xjobs=24 -V -# -v |
cc |
CC |
f90 |
cc f90 |
410.bwaves: | basepeak = yes |
416.gamess: | -fast -xcache=128/64/2:6144/256/12 -xpagesize=4M -xipo=2 -xprefetch_level=2 -fma=fused |
434.zeusmp: | basepeak = yes |
437.leslie3d: | -fast -xcache=128/64/2:6144/256/12 -xpagesize=4M -xprefetch_level=3 -qoption cg -Qlp=1 -qoption cg -Qlp-fa=0 -qoption cg -Qlp-fl=1 -qoption cg -Qlp-av=448 -qoption cg -Qlp-t=4 -xprefetch=latx:3.5 |
459.GemsFDTD: | -fast -xcache=128/64/2:6144/256/12 -xpagesize=4M -fsimple=1 -xprefetch_level=2 -fma=fused -xprefetch=latx:2 |
465.tonto: | -fast -xcache=128/64/2:6144/256/12 -xpagesize=4M -xipo=2 -xprefetch=latx:12 -lfast |
435.gromacs: | -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast(cc) -fast(f90) -xcache=128/64/2:6144/256/12 -xpagesize=4M -xipo=2 -xinline= -xarch=generic -xchip=generic -fsimple=0 -fma=fused |
436.cactusADM: | basepeak = yes |
454.calculix: | basepeak = yes |
481.wrf: | basepeak = yes |