POV-Ray : Documentation : 3.5.7.5 Run Benchmark
  POV-Ray 3.6 Documentation Online View  
3.5.7.4 Run Demo   3.5.7.6 File Queue

3.5.7.5 Run Benchmark

This command instructs POV-Ray for Windows to run the standard POV-Ray benchmark file, using a given set of options.

The file itself and the options used are built into the program and may not be changed (this includes the resolution). Any options specified on the command-line will be ignored.

Be aware that during the render, there is no image display, and furthermore, no output file is produced. The only purpose of the benchmark command is to determine how long it takes to render the standard file with the specific options used.

Immediately before POV-Ray for Windows starts writing the render information to the message display, it will output a line telling you what version of the official benchmark is being used.

If you choose to quote your benchmark results publicly, it is essential that you include the benchmark version and the exact version of POV-Ray that you ran it on. Changes to the benchmark file itself, and optimizations to POV-Ray, may make comparisions between different versions of the benchmark or different versions of POV-Ray meaningless, especially if it is CPU performance that you are testing.

Note that, as of the time of writing, running this benchmark takes a long time ! On a 2ghz Intel Pentium 4™, for example, the benchmark takes approximately one hour. (The POV-Team is looking forward to the day when that figure seems slow ;)

The actual benchmark file used is built-in to the program; it is not dependent on the benchmark.pov file which is included in the scenes\advanced directory. However, if you want to run the benchmark manually, or on another platform, you may use that file (just make sure it is the same version). The file also contains the recommended command-line parameters to be used with the render.

The benchmark command is most useful on NT-based versions of Windows (this includes Windows XP). This is because these operating systems provide a means for POV to determine the actual CPU time used by itself. Since Windows is a multi-tasking operating system, other processes (including the operating system itself) take time away from POV-Ray while it is rendering. In terms of benchmarking a processor, the elapsed time that a render takes is not really relevent - it is the CPU Time that really matters. Remember that the purpose of this benchmark is not to determine how long it takes to render a scene (in which case you care about elapsed time), but to determine the performace of a CPU. That is, it is not intended to be a test of POV-Ray itself, but of the architecture that it is running on.

If you are running Windows 95/98 or ME then you will only be able to get elapsed (clock) time, which is not as accurate. In this case you should make sure that no other applications are running during the render (this includes screensavers and also background apps such as virus checkers), and also you should increase the render priority to the highest level.

If you are running an NT-based operating system, after the render a dialog box will be displayed showing the average PPS (pixels per second) that the render took. Larger numbers are better (indicative of a faster and/or more efficient CPU). We recommend using this number when quoting results.

In case you are interested, here are the results the author of POV-Ray for Windows got on the two machines on which he ran the benchmark v1.01 on (note that this was with POV-Ray for Windows v3.5 final release). The operating system in both cases was Windows 2000 SP2.
2.0 GHz Intel Pentium™ 4
256 MB RDRAM
1.4 GHz AMD Athlon™
1024 MB DDR RAM
54.04 PPS 53.11 PPS

For reference, here are the standard options supplied to the renderer as of the standard benchmark version 1.01 (be sure to check that this is the one you are using if you want to use the below options, as it is possible for them to have been changed without this document being updated).

All_Console=Off
Antialias_Depth=3
Antialias=On
Antialias_Threshold=0.3
Bits_Per_Color=8
Bounding=On
Bounding_Threshold=3
Buffer_Output=Off
Buffer_Size=0
Clock=0
Continue_Trace=Off
Create_Histogram=Off
Cyclic_Animation=Off
Debug_Console=On
Display=Off
Display_Gamma=1.0
Draw_Vistas=Off
End_Column=1
End_Row=1
Fatal_Console=On
Fatal_Error_Command=
Fatal_Error_Return=I
Field_Render=Off
Final_Clock=1
Final_Frame=1
Height=384
Histogram_Name=
Histogram_Grid_Size=0.0
Initial_Clock=0
Initial_Frame=1
Include_Header=
Jitter_Amount=1
Jitter=On
Light_Buffer=On
Odd_Field=Off
Output_Alpha=Off
Output_File_Name=
Output_File_Type=s
Output_To_File=Off
Palette=3
Pause_When_Done=Off
Post_Frame_Command=
Post_Frame_Return=I
Post_Scene_Command=
Post_Scene_Return=I
Preview_End_Size=1
Preview_Start_Size=1
Pre_Frame_Command=
Pre_Frame_Return=I
Pre_Scene_command=
Pre_Scene_Return=I
Quality=9
Remove_Bounds=On
Render_Console=On
Sampling_Method=1
Split_Unions=Off
Start_Column=0
Start_Row=0
Statistic_Console=On
Subset_End_Frame=1
Subset_Start_Frame=1
Test_Abort_Count=0
Test_Abort=Off
User_Abort_Command=
User_Abort_Return=I
Verbose=On
Version=3.5
Video_Mode=0
Vista_Buffer=On
Warning_Console=On
Width=384
3.5.7.4 Run Demo   3.5.7.6 File Queue


Copyright 2003-2004 Persistence of Vision Raytracer Pty. Ltd.