Quote - PJ's and Connie, thanks for the links. The response of shadow_29 at TechSpot was very detailed and eye-opening. PJ, lol, I was conversing with my son last night and he told me the 7's were dead and the 8's were great! I wonder if those are only gamer cards, though. More research on hardware later.

:: og ::

 

Well, here's what I deal with...
At the office, the terrain modelling that we do is all triangulation of point clouds containing couple milion points each... the meshes I handle end up having several milion triangles. 3-5 milion on average. And I get to do almost no rendering at all, it;s all calculations and massing models.
Without 4 GIG of ram and couple of quadro series of workstation cards I couldn't lift a finger. Now, that's the higer end extreme, that you may not ever reach. 
That also depends on how body studio handles importing. What I've seen is that the imports aren't exactly sparing on the triangles since each animation keyframe appears to become it's own set of OBJ's. (if I remember right) I know pomax does for sure, since I used it more recently. I believe body studio does something very similar.

Quadro's have great capability for rendering stills and animations where you are rendering each frame and creating an avi or similar. It pretty much screams through it, but only with programs that are able to take advantage of it's features (like max). 

Where gaming cards seem to have an edge is in real time 3D graphics, where seeing a fully rendered low poly scene in a flash is important for gameplay, and testing the gameplay. 
In gameplay you often only see a portion of a scene, and it's effective loading gets done between the gaming engine, system power and graphics card power (to put it rather simply).

When a level modeller is creating entire model, and going through handling several levels of detail, building it all together in Max, this instant loading and rendering of a small portion fo a scene is not nearly as helpful as the ability to smoothly handle a number of triangles, as vectors or as massing model.

Having said all that, I'll give you an actual example: Working on a poser scene with couple of characters in it (500,000 triangles max) at the office with a 3GHZ-4gig system and 2 quadro cards in it, or at home on a2.5GHZ,  2 gig system and a 2-year old nvidia 6800GT, I can't tell a difference in making models and handling scenes. It's not significantly smoother or faster at the office. 

But, taking one of my work related projects with several milion triangles in it, and working on it at home, I can't do at all. Maximum I can handle at home is a project with 1.2 mil triangles before I crash and burn. At work, the largest one I had so far is 7 mil, and didn't crash. I did something silly trying to load couple of versions of a 7 mil triangle surfaces at onece, and did crash... so the limit is somwehere between 7 and 14 milion.

I haven't done a lot of animation comparisons between the two. When I need to render architectural or similar presentations, I set up several of the office and home computers into a rendering farm, and let them crunch at night. Often we create presentations that have up to 20-30 min worth of drive by's, and flyovers that take several days to render, so I haven't concerned myself with speed improvements less then 50% at the time.

If I can get more output from a two little more of the middle of the road machines then from one super duper latest one, I get the two middle of the road. The prices for the latest and greatest are often so jacked up that you can often get two for one.

Now, if I were practicing quake or similar gameplay for competition, or bragging rights, then I probably would get a $4000 or so gaming powerhouse.

Whenever I have my way, I get the machines from theese guys: http://www.xicomputer.com/

Hi, my namez: "NO, Bad Kitteh, NO!"  Whaz yurs?
BadKittehCo Store  BadKittehCo Freebies and product support