So we just got new Boxx machine at work with 20 cores and the quadro K6000.
First think i did was to try a sphere and divide it as much as i could.
Surprisingly the quadro started slowing down to 10fps in the view port at 20 millions polygons, while my gtx titan goes up to 80 millions polygons at 10 fps.

i tried different drivers for the quadro k6000 but i could never go above 20 millions polys

the Boxx is windows 8.1
64 gb ram
20 cores xeon
quadro k6000
Drivers are up to date.

Does anyone has a solution or explanation as to why it is performing so bad.

thank you

i tried to open the 20 million sphere in gplay.exe and it performs ridiculously well there, also in c4d the viewport is very smooth with the sphere.

Could it be houdini UI that create problem with performance of the quadro?

update:

when going to obj level the viewport go back to excellent speed
also packing the sphere at sop level get the viewport to good speed.

So my question is why at sop level the viewport has trouble displaying the high density object while not at obj level.

in anycase at sop level my titan out performs the quadro as for now

Not sure why the Quadro is slower than the Titan, but, Obj level will always out performed Sop level as you can do more stuff with the geo at obj level.

i discovered something interesting

Geometry information “always on”or “selected only” greatly impact view port peformance.

on a 20 million poly sphere i get 21 fps without geometry information and down to 3 fps when geometry information is on.

final conclusion:

i did test again without the geometry info display and the quadro K6000 performs better than the gtx titan.

the issue was definitely that info option.

final result are:
20 fps on a 80 million polygons for the quadro k6000
15 fps on a 80 million polygons for the GTX titan