Quadro K5000: Amy opinions or tests?

[trick]

I'm speccing a new machine and - as usual - don't know what to do with the graphics.

Most important is MAX2013 working speed, both wireframe and shaded (incl. new Max2013 display moded), then the speed with VRayRT and finally working speed with Lumion3D. Final render speed with Lumion3D is not that important since the scenes can be transferred to a machine with GTX680/4GB.

I'm looking for the Quadro cards since they are more stable, quiet and cool than their GeForce equivalents. I can't find any specific K5000 tests and wonder what the exact advantages are with this Keppler card above the Fermi 4/5/6000 series.

[Morbid Angel]

Not sure about stable or quiet. We have quadro 4000 in the office and when those things go full on its notably loud. Also performance wise its a tad better then geforce, but certainly not worth the price.

[cheerioboy]

Yeah I was drooling over the k5000 for a while. But I just purchased a 680GTX after a co-worker recommended it over quadros. So we'll see how it handles max/vray/etc.

I always thought quadros were better but apparently they're not worth the higher cost?

[Morbid Angel]

on the performance side they handle wireframes and vertices better then geforces, but they have half the cores they modern gforce so gpu rendering is really slower.

[mario74]

i changed the quadro 4000 into a gtx 680 a month ago - in the hope it will change anything within max 2013. my models are mostly 10+ mio polys and the workflow is with the gtx the same. if i turn everything on it allows me to move there with 1.2 up to 1.5 fps... simply unable to work with both of them!

autodesk-center works with quadro 5000, this one is faster than my gtx 680. we run a test and this is for sure, but we talking about peanuts here, both are very close in performance with nitrous driver.

[Morbid Angel]

did you try in max 2012, i read that there were some viewport issues in 2013...

[mario74]

its quite off topic, but i will try to explain

we considered to give vray a try (this was end of year 2011) - after a view tests we decided do change a part of our workflow - we needed 3ds max also. (we never used this software before) but it was not possible to get version of 2012, so i cant compare now.

i am sure, it took more than 30 hours of my life on phone with autodesk, cause max 2013 is not usable with heavy scenes. its not the number of polygones, its more the number of parts in scene. anyway, even autodesk-center gave up because our scenes are too heavy to handle within max 2013... try this, try that, reduce number of parts and add meshes to each other...

i started to love vray's proxys, but they are not usefull when you have to set your camera "like you feel" and you cant see where you are and what is in front of you.

and this brings me straight to RT, that's why i ordered a gtx - and for a preview its quite nice. RT sometimes leave my desktop without any emotion, but for me it is the direction to go!

[trick]

I know that current GeForce cards are faster and still way more cheaper than Quadro equivalents. But still prefer Quadro's because they are more quite, stable and cool which results in far better (read stable) systems, especially dual xeon's. This is from experience! I even have a seperate GTX680 system for doing Nitrous output (and other non-MAX stuff), but when it comes to modelling, which comprises a large part of my workflow, I farly prefer the Quadro system.

But I still would like to know what are the advantages of the K(eppler) Quadros over the non-K ones ?

[Franx]

I know that current GeForce cards are faster and way more expensive than Quadro equivalents. ..

Geforces are actually about 3 to 4 times cheaper than their Quadro equivalents.

[trick]

Geforces are actually about 3 to 4 times cheaper than their Quadro equivalents.

Stupid me !! Of course: changed text !!!!

[mario74]

But I still would like to know what are the advantages of the K(eppler) Quadros over the non-K ones ?

as far as i know, the keppler you can get now, cant swap memory. that means on big modells they will crash system or you need a second one. cause of this reason we did not order a tesla.

[trick]

as far as i know, the keppler you can get now, cant swap memory. that means on big modells they will crash system or you need a second one. cause of this reason we did not order a keppler.

Is this typical for Keppler ? Because on my GTX680's this never happened !

[mario74]

that belongs only to the tesla ones. sorry, i should be more precise

[danio]

For what its worth, if heat and noise is a consideration, I run a watercooled 3gb 580 and love it. Very fast and stable even in big Arch viz scenes and very cool running. Bumping my voltage to 1.013 seems to smooth things out even a bit more.