Tim Cook wrote:
> Thorsten Froehlich wrote:
> 
>> Well, the POV-Team has repeatedly and very clearly stated that one 
>> cannot compare the benchmark results between different versions of 
>> POV-Ray.  And certainly a beta version with plenty of bugs cannot be 
>> compared to an older final version of POV-Ray.  As such, the 
>> comparison about POV-Ray 3.6 release and POV-Ray 3.7 beta says 
>> absolutely nothing about the speed increase.
> 
> 
> It might not say anything about what the software is doing, let alone 
> that it has the same speed increase for every image, but it *does* say 
> that the image in question *was* created in 1/10th the time.  That's 
> saying absolutely something about the speed increase in one specific 
> case.  Which is still impressive.  ;)

You missed my point: The image created is not identical due to the mentioned 
bugs, and some features are disabled.  Hence, the results of comparing 3.6 
release with 3.7 beta just have *no* validity whatsoever.

And believe me, I know very well what the real speed of POV-Ray 3.7 compared 
to 3.6 currently is.  The benchmark is still 10% slower than rendered 3.6 
(and that compares the Visual C builds), and there is no magic in there for 
the multiprocessor case - you get four times the speed of a single threaded 
render, not more.

	Thorsten

Post a reply to this message