Polyhedron Benchmarks (AMD+Windows Sept 2015)

Win64 on AMD Processor

Absoft
15.0.2		 Absoft(AP)
15.0.2		 FTN95
7.20		 Intel
16.0		 Intel(AP)
16.0		 Lahey-GNU
Lassen-4.93		 NAG
6.0		 PGI
15.7
AC 7.65 7.82 15.76 6.92 6.97 8.46 10.19 8.65
AERMOD 19.02 19.85 33.43 17.40 17.49 28.22 24.98 17.99
AIR 6.77 3.18 13.83 4.89 2.44 7.54 7.70 11.52
CAPACITA 37.56 32.29 65.03 37.50 37.18 44.56 49.29 31.90
CHANNEL2 144.55 116.56 363.63 145.95 111.64 151.41 222.53 141.77
DODUC 27.77 28.40 49.45 21.27 21.45 29.75 30.34 26.88
FATIGUE2 81.96 86.98 422.66 75.86 75.84 44.01 128.10 173.18
GAS_DYN2 102.38 103.54 560.43 100.13 99.83 230.00 200.36 130.14
INDUCT2 96.23 69.85 601.08 111.23 44.56 96.88 150.97 167.40
LINPK 9.42 9.28 10.70 9.70 9.73 10.33 10.22 10.30
MDBX 12.47 11.60 24.09 10.62 10.78 11.51 12.50 14.67
MP_PROP_DESIGN 137.74 42.24 692.06 101.05 32.51 282.85 286.72 138.63
NF 18.31 16.79 28.32 14.04 14.04 16.83 19.36 13.91
PROTEIN 31.53 30.73 56.31 31.78 31.86 32.46 32.37 35.48
RNFLOW 17.51 21.29 35.90 18.90 20.11 22.34 24.30 24.25
TEST_FPU2 94.39 89.87 192.98 85.44 80.59 102.44 118.30 89.20
TFFT2 135.46 133.22 168.52 135.93 135.99 89.08 137.72 138.16
Geometric Mean 35.72 30.77 82.79 32.81 27.53 39.40 46.59 40.93
Compiler Switches
Absoft
 af90 -m64 -O4 -speed_math=10 -fast_math-march=barcelona -xINTEGER -stack:0×30000000
Absoft (autoparallel)
 af90 -m64 -O5 -speed_math=10 -fast_math -march=barcelona -xINTEGER -stack:0×30000000
FTN95 ftn95 /p6 /optimize (slink was used to increase the stack size)
Intel ifort /fast /link /stack:64000000
Intel (autoparallel) ifort /fast /Qparallel /link /stack:64000000
Lahey-GNU lgf -64 -ofast -unroll -wpo -stack 64000000
NAG nagfor -abi=64 -O4 -s -v -V
PGI pgf90 -V -fastsse -Munroll=n:4 -Mipa=fast,inline
Notes
All figures are execution times in seconds – measured on a machine with an AMD Phenom II X4 955 processor (3.2 GHz), with 8 GBytes memory, running Windows 7 64-bit. Each figure is the average over at least 10 runs (many more for some). Measurement error is typically <1%. Green cells highlight figures within 10% of the fastest. Red cells indicate figures which are more than 150% of the fastest. So far as possible, we have used the compiler switches which give the best overall results. We have not attempted to tune individual benchmarks, and, in particular cases, different switch settings may give better results. We have created and used 64 bit executables where possible, and 32 bit executables where the compiler does not offer a 64 bit option. The settings used for the Absoft and Intel compilers enable autoparallelization. Autoparallelization settings are not used on any other compilers because we found that they produced no significant performance benefits on this benchmark set. Thanks are due to Jos Bergervoet for permission to use his CAPACITA benchmark, to Quetzal Associates for permission to use their CHANNEL, FATIGUE, GAS_DYN, INDUCT, PROTEIN and RNFLOW benchmarks, to David Frank for his TEST_FPU benchmark, to Anthony Falzone for the use of MP_PROP_DESIGN, and to Ted Addison of McVehil-Monnett Associates for permission to use AERMOD, an air quality model used by the US Environmental Protection Agency. All the benchmarks have been modified slightly to fit into our benchmarking harness.