> POV reports that the "Render" -> "Run Benchmark" menu option takes
> about one hour to execute on a 2GHz Pentium 4 machine. Well, you might
> want to revise that warning dialog, just before you release the final
> version of POV-Ray. I just ran the official benchmark from that menu,
> and it rendered in only 46 minutes, 26 seconds.

I got nearly the same with Athlon TBird 1.4GHz (46:57)

I looked up the help for benchmark (Windows specific: 1.10.7.1.5), and it
says, that the author of Povray for Windows (That would be Chris Cason) ran
the benchmark with the final release of 3.5:
"(note that this was with POV-Ray for Windows v3.5 final release)"
is there a "not" missing?

The average PPS of the mentioned P4 2000: 54.04 PPS leads exactly to one hour
: (512*384)/54.04=3638sec, 3600 would be an hour.

The problem is, that povray isn't running the benchmark with 512*384, but
with 384*384 instead, this can be seen in the message-window:

<quote from message-stream>
Running standard POV-Ray benchmark version 1.01

Persistence of Vision(tm) Ray Tracer Version 3.5.beta.RC6 win32 (.icl)
  This is a time-limited beta test version.  No re-distribution
   of any sort is permitted.
...
...

Output Options
  Image resolution 384 by 384 (rows 1 to 384, columns 1 to 384).
</quote from message-stream>


(Taking this into account, those Athlons would take 1 hour and 2.5 minutes.
-> not too fast anymore)

There's one thing to add:

This benchmark generates a warning:

<quote from message-stream>
Warning: The maximum gradient found was 2.626, but max_gradient of the
isosurface was set to 2.400. The isosurface may contain holes!
Adjust max_gradient to get a proper rendering of the isosurface.
</quote from message-stream>

Maybe this is right, I don't know (perhaps the benchmark is tweaked to look
good and be "fast"; at least not unnecessarily slow...)

cukk

Post a reply to this message