|
![](/i/fill.gif) |
Ken wrote:
>
> Benchmark testing generaly relies on the amount of time it takes
> to render a given scene. If one were to design a scene that was
> parsing intensive rather than render intensive how well would it
> evaluate a systems performance ?
Comment:
- Rendering relies heavily on floating point operations.
- Parsing often relies more heavily on integer operations.
- Parsing can often use a lot of memory, so it can test memory
performance.
- Not that you asked, but lots of includes, big external files, etc. can
test disk performance.
Much depends on the nature of the code being parsed. You could write
code that requires a lot of floating point, tests the branch prediction
in your CPU, overflows cache and has to hit main memory a lot, etc.
I would suggest that benchmarks that don't rely on much extraneous
parsing would be most useful for determining how quickly a system
renders scenes, although a suite of scenes (rather than a single
skyvase.pov) might be useful.
-Mark Gordon
Post a reply to this message
|
![](/i/fill.gif) |