Fax: +44(0)1865 300 232

Programs for Programmers

64 bit Fortran Execution Time Benchmarks - Linux64 on AMD Phenom II

  Absoft
13.0.0
G95
0.93
GFortran
4.7.2
Intel(AP)
14.0
Intel
14.0
Lahey
8.10b
PGI
13.7
Sun
8.6
AC 7.38 14.66 8.69 9.71 9.54 11.56 8.71 29.20
AERMOD 18.07 38.81 28.20 17.21 17.32 18.26 17.33 15.46
AIR 2.51 11.56 6.68 2.23 4.61 4.82 6.79 4.95
CAPACITA 31.35 56.79 43.79 36.65 36.68 42.78 30.39 51.19
CHANNEL2 154.55 468.22 165.53 110.18 163.21 237.35 201.52 158.04
DODUC 27.39 35.31 26.13 22.64 22.35 27.06 25.41 22.88
FATIGUE2 131.92 559.28 141.02 67.24 67.13 162.95 114.35 114.38
GAS_DYN2 94.33 449.04 148.01 117.10 121.91 167.31 155.80 120.97
INDUCT2 40.99 281.32 88.38 45.08 110.58 154.40 200.74 190.58
LINPK 10.75 11.39 10.69 10.58 10.55 11.24 11.22 10.89
MDBX 11.44 13.26 13.05 10.92 10.65 12.40 13.00 11.99
MP_PROP_DESIGN 40.00 711.21 163.14 34.19 105.44 202.11 125.73 209.06
NF 13.60 24.70 12.32 12.16 12.19 18.41 13.00 13.09
PROTEIN 32.07 44.71 31.53 30.74 30.85 48.20 36.03 31.64
RNFLOW 15.57 38.76 24.50 18.68 18.72 25.08 23.20 26.81
TEST_FPU2 83.65 174.64 91.63 84.83 88.87 109.93 93.07 105.66
TFFT2 130.77 137.96 130.48 135.44 131.76 128.27 135.33 128.99
 
Geometric Mean 29.23 73.08 39.65 27.93 33.63 44.47 39.57 42.33

 

Compiler Switches
Absoft

af95 -m64 -O5 -speed_math=10 -march=barcelona -xINTEGER (These settings enable auto-parallelization, see below)

g95 g95 -march=opteron -ffast-math -funroll-loops -O3
gfortran gfortran -march=amdfam10 -ffast-math -funroll-loops -O3
Intel(AP) ifort -O3 -fast -parallel -ipo (These settings enable auto-parallelization, see below)
Intel ifort -O3 -fast -ipo
Lahey lf95 --fast -static -x -
PGI pgf90 -V -fast -Mipa=fast,inline -Msmartalloc -Mfprelaxed -Mstack_arrays
Sun sunf95 -fast -xtarget=native

 

Notes
All figures are Execution Times in Seconds - measured on a machine with an AMD Phenom II X4 955 processor (3.2 GHz), with 4GBytes memory, running CentOS 5.5. 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.

The settings used for the Intel and Absoft 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. 

The NF benchmark uses  "nested factorization", a little known but very effective iterative linear solver for huge finite difference matrices.  A paper describing nested factorization, and comparing it to other methods is available here.

 

Other benchmark sets:

AMD/Windows

Intel/Windows (Sandy Bridge)

Intel/Linux


 

Download Polyhedron Benchmarks