| CPU2006 license: | 19 | Test date: | Jan-2009 |
|---|---|---|---|
| Test sponsor: | Fujitsu | Hardware Availability: | Jul-2009 |
| Tested by: | Sun Microsystems | Software Availability: | Nov-2008 |
| Hardware | |
|---|---|
| CPU Name: | UltraSPARC T2 |
| CPU Characteristics: | |
| CPU MHz: | 1582 |
| FPU: | Integrated |
| CPU(s) enabled: | 8 cores, 1 chip, 8 cores/chip, 8 threads/core |
| CPU(s) orderable: | 1 chips |
| Primary Cache: | 16 KB I + 8 KB D on chip per core |
| Secondary Cache: | 4 MB I+D on chip per chip |
| L3 Cache: | None |
| Other Cache: | None |
| Memory: | 64 GB (16 x 4 GB) |
| Disk Subsystem: | 142 GB Solaris Volume Manager RAID 0 on 4x 72 GB Sun 10K RPM SAS block size 384 KB |
| Other Hardware: | None |
| Software | |
|---|---|
| Operating System: | Solaris 10 10/08 |
| Compiler: | Sun Studio 12 and gccfss V4.2.1 (see additional detail below) |
| Auto Parallel: | No |
| 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 | 63 | 9523 | 89.9 | 9574 | 89.4 | 9571 | 89.5 | 63 | 9523 | 89.9 | 9574 | 89.4 | 9571 | 89.5 |
| 416.gamess | 63 | 21500 | 57.4 | 21484 | 57.4 | 21495 | 57.4 | 63 | 20761 | 59.4 | 20765 | 59.4 | 20767 | 59.4 |
| 433.milc | 63 | 7066 | 81.8 | 7067 | 81.8 | 7064 | 81.9 | 63 | 7066 | 81.8 | 7067 | 81.8 | 7064 | 81.9 |
| 434.zeusmp | 63 | 8186 | 70.0 | 8133 | 70.5 | 8164 | 70.2 | 63 | 8186 | 70.0 | 8133 | 70.5 | 8164 | 70.2 |
| 435.gromacs | 63 | 7806 | 57.6 | 7818 | 57.5 | 7871 | 57.1 | 63 | 7679 | 58.6 | 7677 | 58.6 | 7713 | 58.3 |
| 436.cactusADM | 63 | 12977 | 58.0 | 13023 | 57.8 | 13005 | 57.9 | 63 | 12664 | 59.4 | 12622 | 59.6 | 12614 | 59.7 |
| 437.leslie3d | 63 | 10422 | 56.8 | 10384 | 57.0 | 10372 | 57.1 | 48 | 7569 | 59.6 | 7553 | 59.7 | 7543 | 59.8 |
| 444.namd | 63 | 7729 | 65.4 | 7713 | 65.5 | 7714 | 65.5 | 63 | 7598 | 66.5 | 7633 | 66.2 | 7598 | 66.5 |
| 447.dealII | 63 | 7096 | 102 | 7110 | 101 | 7091 | 102 | 63 | 6805 | 106 | 6787 | 106 | 6772 | 106 |
| 450.soplex | 63 | 7515 | 69.9 | 7500 | 70.1 | 7506 | 70.0 | 63 | 7515 | 69.9 | 7500 | 70.1 | 7506 | 70.0 |
| 453.povray | 63 | 3877 | 86.4 | 3803 | 88.1 | 3806 | 88.1 | 63 | 2989 | 112 | 2942 | 114 | 2923 | 115 |
| 454.calculix | 63 | 11931 | 43.6 | 11950 | 43.5 | 11936 | 43.5 | 63 | 11931 | 43.6 | 11950 | 43.5 | 11936 | 43.5 |
| 459.GemsFDTD | 63 | 12996 | 51.4 | 12993 | 51.4 | 12994 | 51.4 | 63 | 12996 | 51.4 | 12993 | 51.4 | 12994 | 51.4 |
| 465.tonto | 63 | 10891 | 56.9 | 10981 | 56.5 | 10990 | 56.4 | 56 | 8791 | 62.7 | 8812 | 62.5 | 8867 | 62.1 |
| 470.lbm | 63 | 28620 | 30.2 | 28613 | 30.3 | 28627 | 30.2 | 8 | 2139 | 51.4 | 2139 | 51.4 | 2140 | 51.4 |
| 481.wrf | 63 | 9690 | 72.6 | 9700 | 72.5 | 9682 | 72.7 | 63 | 9540 | 73.8 | 9566 | 73.6 | 9544 | 73.7 |
| 482.sphinx3 | 63 | 14916 | 82.3 | 14940 | 82.2 | 14961 | 82.1 | 63 | 14518 | 84.6 | 14512 | 84.6 | 14518 | 84.6 |
Compiler Invocation Notes
Sun Studio compiler patches are available at http://developers.sun.com/sunstudio/downloads/patches/ss12_patches.jsp The tested configuration included patch 124867-08, 124861-09, 124863-08, 127000-06 Peak also uses "GCC for SPARC Systems 4.2.1", which combines gcc with the Sun Code Generator for SPARC systems. It is invoked as "gcc", and accepts source code compatible with GCC 4.2. For more information, including support, see http://cooltools.sunsource.net/gcc/
Submit Notes
A processor set was created using psrset -c 1-63 and the runspec process was placed into the set using psrset -e 1 The config file option 'submit' was used to select specific processors within the set, along with the pbind command.
Operating System Notes
ulimit -s 131072 was used to allow the stack to grow
up to 131072 KB (aka 128 MB). Note that saying "131072"
is preferable to "unlimited", because there is a tradeoff
between space for the stack vs. space for the heap.
/etc/system parameters
autoup=600
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=10
Controls how many seconds elapse between runs of the
page flush daemon, fsflush.
tsb_rss_factor=128
Suggests that the the size of the TSB (Translation Storage Buffer)
may be increased if it is more than 25% (128/512) full. Doing so
may reduce TSB traps, at the cost of additional kernel memory.
The "webconsole" service was turned off using
svcadm disable webconsole
The system had 75 GB of swap space.
Platform Notes
This result was measured on a Sun SPARC Enterprise T5220. All of these are electronically equivalent: - Sun SPARC Enterprise T5120 - Sun SPARC Enterprise T5220 - Fujitsu SPARC Enterprise T5120 - Fujitsu SPARC Enterprise T5220 A SPARC Enterprise 5120 can hold up to 4 disks, and a 5220 can hold up to 8. This system was tested with 4 disks; therefore, this result applies to both the 5120 and the 5220.
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:
| -g -fast -xipo=2 -xpagesize=4M -xprefetch_level=2 -xalias_level=std -xprefetch_level=3 -xprefetch_auto_type=indirect_array_access -M /usr/lib/ld/map.bssalign |
C++ benchmarks:
| -g0 -library=stlport4 -fast -xipo=2 -xpagesize=4M -xprefetch_level=2 -xdepend -xalias_level=compatible -M /usr/lib/ld/map.bssalign |
Fortran benchmarks:
| -g -fast -xipo=2 -xpagesize=4M -xprefetch_level=2 -M /usr/lib/ld/map.bssalign |
Benchmarks using both Fortran and C:
| -g -fast(cc) -fast(f90) -xipo=2 -xpagesize=4M -xprefetch_level=2 -xalias_level=std -xprefetch_level=3 -xprefetch_auto_type=indirect_array_access -M /usr/lib/ld/map.bssalign |
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 (except as noted below):
| CC | |
| 447.dealII: | g++ |
Fortran benchmarks:
| f90 |
Benchmarks using both Fortran and C:
| cc f90 |
Peak Optimization Flags
C benchmarks:
| 433.milc: | basepeak = yes |
| 470.lbm: | -g -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast -xpagesize=4M -M /usr/lib/ld/map.bssalign -xprefetch_level=3 -xipo=2 -xrestrict |
| 482.sphinx3: | -g -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast -xpagesize=4M -M /usr/lib/ld/map.bssalign -xinline= -xprefetch_level=2 -Wc,-Qlp-ol=1 -xrestrict -xalias_level=strong -fsimple=1 -xlinkopt=2 -lfast |
C++ benchmarks:
| 444.namd: | -g0 -library=stlport4 -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast -xpagesize=4M -xdepend -xalias_level=compatible -M /usr/lib/ld/map.bssalign -xprefetch_level=1 -xlinkopt=2 |
| 447.dealII: | -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast -xpagesize=4M -xdepend -Wl,-M,/usr/lib/ld/map.bssalign -xipo=2 -xrestrict -xalias_level=std |
| 450.soplex: | basepeak = yes |
| 453.povray: | -g0 -library=stlport4 -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast -xpagesize=64K -xdepend -xalias_level=compatible -xipo=2 -xrestrict -xlinkopt=2 |
Fortran benchmarks:
| 410.bwaves: | basepeak = yes |
| 416.gamess: | -g -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast -xpagesize=4M -M /usr/lib/ld/map.bssalign -xlinkopt=2 |
| 434.zeusmp: | basepeak = yes |
| 437.leslie3d: | -g -fast -xpagesize_heap=4M -xpagesize_stack=64K -M /usr/lib/ld/map.bssalign -xprefetch_level=3 -xprefetch=latx:1.6 -qoption cg -Qlp=1 -qoption cg -Qlp-fa=0 -qoption cg -Qlp-fl=1 -qoption cg -Qlp-av=448 -qoption cg -Qlp-t=4 |
| 459.GemsFDTD: | basepeak = yes |
| 465.tonto: | -g -fast -xpagesize=4M -M /usr/lib/ld/map.bssalign -xipo=2 -lfast |
Benchmarks using both Fortran and C:
| 435.gromacs: | -g -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast(cc) -fast(f90) -xpagesize=4M -M /usr/lib/ld/map.bssalign -xipo=1 -xinline= -xarch=generic -xchip=generic -fsimple=0 |
| 436.cactusADM: | -g -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast(cc) -fast(f90) -xpagesize=4M -M /usr/lib/ld/map.bssalign -xipo=2 -fsimple=1 -xlinkopt=2 |
| 454.calculix: | basepeak = yes |
| 481.wrf: | -g -xprofile=collect:./feedback(pass 1) -xprofile=use:./feedback(pass 2) -fast(cc) -fast(f90) -xpagesize=4M -M /usr/lib/ld/map.bssalign -xlinkopt=2 |