Baseline optimization flags:
C programs: -openmp -O3 -IPF_fp_relaxed -ipo -ansi -ansi_alias -auto_ilp32 (ONESTEP)
Fortran programs: -openmp -O3 -IPF_fp_relaxed -ipo (ONESTEP)
OpenMP runtime library libguide.a statically linked
Portability Flags:
318.galgel_m: -FI -132
Extra Flags:
330.art_m: -DINTS_PER_CACHELINE=32 -DDBLS_PER_CACHELINE=16
User environment:
OMP_NUM_THREADS 32
limit stacksize 64000
KMP_STACKSIZE 31M
KMP_LIBRARY TURNAROUND
OMP_DYNAMIC FALSE
KMP_SCHEDULE static,balanced
Peak optimization flags:
310.wupwise_m: basepeak=true
312.swim_m: basepeak=true
314.mgrid_m: basepeak=true
316.applu_m: basepeak=true
318.galgel_m: basepeak=true
320.equake_m: -openmp -O3 -IPF_fp_relaxed -ipo -ansi -ansi_alias -auto_ilp32 (ONESTEP)
324.apsi_m: -openmp -O3 -IPF_fp_relaxed -ipo (ONESTEP)
326.gafort_m: -openmp -O3 -IPF_fp_relaxed -ipo (ONESTEP)
328.fma3d_m: -openmp -O3 -IPF_fp_relaxed -ipo (ONESTEP)
330.art_m: basepeak=true
332.ammp_m: basepeak=true
Required alternate sources:
Add critical region around update of linked list in parallel loop.
Approved src.alt available as ompm-purdue1-20040324.tar.gz
Used for 330.art_m, base and peak.
Peak sources:
SPEC OMPL2001 source for 64bit systems modified for SPEC OMPM2001.
Available as ompl src.alt in SPEC OMP v3.0
Used for 320.equake_m, 324.apsi_m, 326.gafort_m, and 328.fma3d_m.
For all benchmarks threads were bound to cores using the following submit command:
dplace -x2 -cNTM1,0 $command,
where NTM1 is the number of threads minus 1.
This binds threads in order of creation, beginning with the master
thread on core NTM1, the first slave thread on core NTM1-1, and so on.
The -x2 flag instructs dplace to skip placement of the lightweight
OpenMP monitor thread, which is created prior to the slave threads.
For a description of SGI's compiler flags, portability flags, and
system parameters used to generate this result, please refer to the
SGI-20041118-Linux-Intel8.1-IPF.txt file in the flags directory.
|
