We're looking at purchasing new hardware, and are in need of having benchmarks run on our custom software.

FARSITE (Fire Area Simulator) is a program for computing fire spread given a landscape, a fire perimeter, initial fuel moisture conditions, and a weather stream. Farsite computes arrival time of the fire as it spreads, as well as other various fire intensity and spread rate outputs.

We have a new DLL called Farsite.dll, and I've put together a couple of quick benchmark runs to gather run times on. At this point the FarsiteDLL is single threaded but we will also be running multiple simultaneous instances of Farsite.

The benchmark are very simple to run. Simple download the appropriate package, extract to a directory, and run the two batch files.

RunBenchMark.bat - Launches single threaded TestFarsite.exe.
RunMultiBench.bat - Launches MultiFarsite.exe which creates a Farsite run on each processor and runs them to completion.

The output from each is an arrival time grid and a timings text file which contains run time information. Please zip up all of the .txt files produced by the runs and email them to me or simply post your results here. The readme.txt file contains the email address I would like results submitted to: stubrittainATgmail.com

Here's the links to the benchmark packages:
32 bit Windows Version: http://sbrittain.net/Farsite/BenchFarsitex86.zip
64 bit Windows Version: http://sbrittain.net/Farsite/BenchFarsitex64.zip

I would also like to know 1) If the test machine is overclocked and if so to what processor speed and 2) If hyperthreading was enabled/disabled when running the 'Multi' benchmark.

Thanks in advance to all who run this!

Here's a couple of sample outputs:
Quote Originally Posted by 8x Opteron 8350 - Stock
Farsite BenchMark Results
System Info:
WinMajor: 5, WinMinor: 2, Platform: 2, Build: 3790
Processor: Quad-Core AMD Opteron(tm) Processor 8350
Vendor: AuthenticAMD
Family: 16
Model: 2
Stepping: 3
Speed (MHz): 2009


Simulation Info:
Number Processors Available: 32
Landscape File: 150239.lcp
Landscape Resolution: 30.0
Landscape Size: 371 x 371
Farsite Start Time: 07 31 0600
Farsite End Time: 08 03 1300
Total Sim Time: 4740.000000
Ignition File: ignits.shp
Barriers File: None
Time Step: 60.000000
Distance Resolution: 30.000000
Perimeter Resolution: 60.000000
Enable Spotting: No
Spot Probability: 5.000000
Spot Ignition Delay: 5.000000
Foliar Moisture: 100.0000
Enable Crowning: Yes
Crown Fire Method: Finney
Crown Density NOT LINKED to Crown Cover
Dist Check Method: Fire Level
Acceleration: On
Backing Spread: Calculated from Elliptical Dimensions
Simulation Started: 7/31 6:00
Simulation Ended: 08/03 13:00
Elapsed Time: 03 07:00
Restore Ignitions: No
Preserve Inactive Enclaves: Yes
Burn Periods: 0

Total Farsite Run Time: 191.068 Seconds
Quote Originally Posted by i7 790 - Stock
Farsite BenchMark Results
System Info:
WinMajor: 6, WinMinor: 1, Platform: 2, Build: 7600
Processor: Intel(R) Core(TM) i7 CPU 960 @ 3.20GHz
Vendor: GenuineIntel
Family: 6
Model: 26
Stepping: 5
Speed (MHz): 3238


Simulation Info:
Number Processors Available: 8
Landscape File: 150239.lcp
Landscape Resolution: 30.0
Landscape Size: 371 x 371
Farsite Start Time: 07 31 0600
Farsite End Time: 08 03 1300
Total Sim Time: 4740.000000
Ignition File: ignits.shp
Barriers File: None
Time Step: 60.000000
Distance Resolution: 30.000000
Perimeter Resolution: 60.000000
Enable Spotting: No
Spot Probability: 5.000000
Spot Ignition Delay: 5.000000
Foliar Moisture: 100.0000
Enable Crowning: Yes
Crown Fire Method: Finney
Crown Density NOT LINKED to Crown Cover
Dist Check Method: Fire Level
Acceleration: On
Backing Spread: Calculated from Elliptical Dimensions
Simulation Started: 7/31 6:00
Simulation Ended: 08/03 13:00
Elapsed Time: 03 07:00
Restore Ignitions: No
Preserve Inactive Enclaves: Yes
Burn Periods: 0

Total Farsite Run Time: 91.812 Seconds
Quote Originally Posted by Xeon X5450 - Stock
Farsite BenchMark Results
System Info:
WinMajor: 5, WinMinor: 2, Platform: 2, Build: 3790
Processor: Intel(R) Xeon(R) CPU X5450 @ 3.00GHz
Vendor: GenuineIntel
Family: 6
Model: 23
Stepping: 6
Speed (MHz): 2993


Simulation Info:
Number Processors Available: 8
Landscape File: 150239.lcp
Landscape Resolution: 30.0
Landscape Size: 371 x 371
Farsite Start Time: 07 31 0600
Farsite End Time: 08 03 1300
Total Sim Time: 4740.000000
Ignition File: ignits.shp
Barriers File: None
Time Step: 60.000000
Distance Resolution: 30.000000
Perimeter Resolution: 60.000000
Enable Spotting: No
Spot Probability: 5.000000
Spot Ignition Delay: 5.000000
Foliar Moisture: 100.0000
Enable Crowning: Yes
Crown Fire Method: Finney
Crown Density NOT LINKED to Crown Cover
Dist Check Method: Fire Level
Acceleration: On
Backing Spread: Calculated from Elliptical Dimensions
Simulation Started: 7/31 6:00
Simulation Ended: 08/03 13:00
Elapsed Time: 03 07:00
Restore Ignitions: No
Preserve Inactive Enclaves: Yes
Burn Periods: 0

Total Farsite Run Time: 124.156 Seconds
I am primarily interested in the results from RunBenchMark.bat (either x64 or x86). If you have a machine with hypethreading enabled I would be very interested in the RunMultiBench.bat results files as I'm trying to evealuate the cost/benefits of using hyperthreded machines.

For instance, here's the results of the i7 - 790 system above when running the Multi bench:
Quote Originally Posted by MultiBench on i7 790 Mutli w/Hyperthreading - Stock
Farsite BenchMark Results
System Info:
WinMajor: 6, WinMinor: 1, Platform: 2, Build: 7600
Processor: Intel(R) Core(TM) i7 CPU 960 @ 3.20GHz
Vendor: GenuineIntel
Family: 6
Model: 26
Stepping: 5
Speed (MHz): 3238


Simulation Info:
Number Processors Available: 8
Landscape File: 150239.lcp
Landscape Resolution: 30.0
Landscape Size: 371 x 371
Farsite Start Time: 07 31 0600
Farsite End Time: 08 03 1300
Total Sim Time: 4740.000000
Ignition File: ignits.shp
Barriers File: None
Time Step: 60.000000
Distance Resolution: 30.000000
Perimeter Resolution: 60.000000
Enable Spotting: No
Spot Probability: 5.000000
Spot Ignition Delay: 5.000000
Foliar Moisture: 100.0000
Enable Crowning: Yes
Crown Fire Method: Finney
Crown Density NOT LINKED to Crown Cover
Dist Check Method: Fire Level
Acceleration: On
Backing Spread: Calculated from Elliptical Dimensions
Simulation Started: 7/31 6:00
Simulation Ended: 08/03 13:00
Elapsed Time: 03 07:00
Restore Ignitions: No
Preserve Inactive Enclaves: Yes
Burn Periods: 0

Total Farsite Run Time: 146.438 Seconds
So when running on all logical cores my individual instance time went from 91 secs to 146 seconds. Tests have shown run times to be about the same on non-hyperthreaded machines.

Thanks in advance for running this!