|
|
OMPM2001 Result Copyright © 1999-2008 Standard Performance Evaluation Corporation |
|
IBM IBM HS22 Blade servers (Intel Xeon X5570, 2.93 GHz) |
|
| SPEC license # HPG3440A | Tested by: Indiana University | Test site: -- | Test date: Apr-2009 | Hardware Avail: Apr-2009 | Software Avail: Jan-2009 |
| Benchmark | Reference Time |
Base Runtime |
Base Ratio |
Peak Runtime |
Peak Ratio |
![]() |
|---|---|---|---|---|---|---|
| 310.wupwise_m | 6000 | 84.3 | 71175 | 83.1 | 72213 | |
| 312.swim_m | 6000 | 160 | 37393 | 161 | 37347 | |
| 314.mgrid_m | 7300 | 210 | 34777 | 208 | 35081 | |
| 316.applu_m | 4000 | 126 | 31718 | 127 | 31416 | |
| 318.galgel_m | 5100 | 117 | 43597 | 116 | 43825 | |
| 320.equake_m | 2600 | 54.4 | 47832 | 50.1 | 51888 | |
| 324.apsi_m | 3400 | 86.1 | 39508 | 85.9 | 39601 | |
| 326.gafort_m | 8700 | 164 | 53144 | 169 | 51553 | |
| 328.fma3d_m | 4600 | 150 | 30680 | 141 | 32707 | |
| 330.art_m | 6400 | 64.7 | 98915 | 59.9 | 106797 | |
| 332.ammp_m | 7000 | 249 | 28149 | 232 | 30210 | |
| SPECompMbase2001 | 43657 | |||||
| SPECompMpeak2001 | 44792 | |||||
|
|
||||||||||||||||||||||||||||||||||||||||||||
| Notes / Tuning Information |
|---|
ulimit -s unlimited
Removes limits on the maximum size of the automatically-
extended stack region of the current process and each
process it creates.
Compiler flags for base level optimization
COPTIMIZE : -O3 -xhost -ipo -no-prec-div -unroll-loops0 -openmp
FOPTIMIZE : -O3 -xhost -ipo -no-prec-div -unroll-loops0 -openmp
F77OPTIMIZE : -O3 -xhost -ipo -no-prec-div -unroll-loops0 -openmp
Environment:
KMP_AFFINITY=disabled
controls the binding of OpenMP threads to the physical processing units
KMP_SCHEDULE=static,balanced
used to fine tune the load balancing of parallel loops that are
statically scheduled under OpenMP with no chunk size specification
KMP_BLOCKTIME=infinite
Sets the time, in milliseconds, that a thread should wait,
after completing the execution of a parallel region, before sleeping.
KMP_LIBRARY=throughput
Selects the OpenMP run-time library
KMP_STACKSIZE=31m
Sets the number of bytes to allocate for each parallel thread to use as
to use as its provate stack
OMP_NESTED=TRUE
Enables (TRUE) or disables (FALSE) nested parallelism.
OMP_DYNAMIC=FALSE
Enables (true) or disables (false) the dynamic adjustment of the number of threads.
OMP_NUM_THREADS=16
Sets the maximum number of threads to use for OpenMP* parallel
regions if no other value is specified in the program itself.
Hyperthreading and turbo mode enabled.
318.galgel_m portability flags:
FFLAGS=-fixed -extend-source 132
Flags for peak level optimization
310.wupwise_m peak flags:
fdo_pre0 = rm -rf ./*.dyn
PASS1_FFLAGS = -prof-gen
PASS2_FFLAGS = -prof-use
PASS1_LDFLAGS = -prof-gen
PASS2_LDFLAGS = -prof-use
312.swim_m peak flags:
fdo_pre0 = rm -rf ./*.dyn
PASS1_FFLAGS = -prof-gen
PASS2_FFLAGS = -prof-use
PASS1_LDFLAGS = -prof-gen
PASS2_LDFLAGS = -prof-use
316.applu_m peak flags:
COPTIMIZE = -O2 -xhost -ipo -no-prec-div -openmp
FOPTIMIZE = -O2 -xhost -ipo -no-prec-div -openmp
F77OPTIMIZE = -O2 -xhost -ipo -no-prec-div -openmp
318.galgel_m peak flags:
ENV_OMP_NUM_THREADS=8
320.equake_m peak flags:
ENV_OMP_NUM_THREADS=8
srcalt:ompl.32
324.apsi_m peak flags:
COPTIMIZE = -O2 -xhost -ipo -no-prec-div -openmp
FOPTIMIZE = -O2 -xhost -ipo -no-prec-div -openmp
F77OPTIMIZE = -O2 -xhost -ipo -no-prec-div -openmp
328.fma3d_m peak flags:
srcalt:ompl.32
fdo_pre0 = rm -rf ./*.dyn
PASS1_FFLAGS = -prof-gen
PASS2_FFLAGS = -prof-use
PASS1_LDFLAGS = -prof-gen
PASS2_LDFLAGS = -prof-use
330.art_mpeak flags:
COPTIMIZE = -O2 -xhost -ipo -no-prec-div -openmp
FOPTIMIZE = -O2 -xhost -ipo -no-prec-div -openmp
F77OPTIMIZE = -O2 -xhost -ipo -no-prec-div -openmp
332.ammp_m peak flags:
COPTIMIZE = -O2 -xhost -openmp
FOPTIMIZE = -O2 -xhost -openmp
F77OPTIMIZE = -O2 -xhost -openmp
For a description of the flags used, please see
Indiana-ic11.0-intel64-linux-flags-file-20090428.html in the flags directory
|
First published at SPEC.org on 20-May-2009
Generated on Wed May 20 18:12:06 2009 by SPEC OMP2001 HTML formatter v1.01