Benchmark patch

As there is currently an interesting thread on the shoutbox re. running vvvv on intel macs using Apple Bootcamp, I’d like to open this thread to, well, inspire the creation of a

vvvv benchmark patch.

I bet many here would like to have a set of test patches to judge how good their system is compared to others (especially that fat Mac Power Book there, or that nifty Mac Intel mini!).
Let’s collect some ideas on what to test:

  • Polygon count / Frame Rate ratio
  • Shader compatibility
  • Video stream playback
  • Multiple Video streams at once playback
  • transferring many Textures from HD to GPU

and on features!

  • automatic upload of test profile to a yet-to-be-designed Benchmark database on vvvv-web.
  • Bar charts. oh yeah. realtime bar charts.

???

I see, that I’m not the first with this idea.
We really should do this!

i guess its not worth testing the ‘nifty mac intel mini’ because of Intel GMA950 graphics processor ;)

same applies to the macbook

i want to suggest the exact same thing, so i just revived this thread.
a benchmark patch (e.g. in girlpower) would be awesome! max already lined out a nice test domain i think. #
could be integrated into girlpower, use default images, could give quantifiable information on the power of your system regarding CPU,GPU and harddisk streaming capabilities! very useful imo…

on a side note, and probably not as easily realized, would be performance demand and scaling figures for vvvv’s nodes - e.g. x cpu cycles per spreadcount and frame, scales with n*log(n). example: perlin vs simplex noise

i want to contribute a little sketch regarding vertex count. the rule is simple.

add vertexes as long as the renderer is able to reach 60 fps.

as each adding consumes fps the benchmark will get slower and slower til the end. maybe someone has a better idea but i think this could work…just takes time (5 min)

my card managed to display 4 million vertexes.

see patch

vertex_benchmark 0.1.v4p (16.4 kB)

just tested the patch on my macbook pro and got 1.04 million vertexes. the benchmark was a lot quicker cos less vertexes to add.

but all in all it draws a picture…difference between geforce 8600gt and 8800 gtx.

due to high steps while adding, the results vary a little bit. the steps could be smaller but then it takes longer.