My work computer just got changed out. I went from a Quadro K1100M to a AMD FirePro M5100. I’ve attached some examples of the difference in quality. IMO, the AMD is a significant step down. Lines are much thicker, and antialiasing is poorer, even at higher sampling rates. Line weights also flicker from thick to thin as you pan/zoom/rotate your views. I believe the Quadro screenshot was taken at 4x, and the FirePro at 8x.
Much less fidelity and poor rendering of fine details with the FirePro. It’s passable, but when working with complex models you really notice the difference.
Correct. This has been discussed and shown many times and is well covered in past threads. For some people the antialiasing is very important. For others, not so much. It’s personal preference. About a year ago, AMD has some pre-release drivers out in limited release that did a good job on AA. They never made it to public release. No idea why.
Which to you think it is better for work in rhino in complex models?
QUADRO are better in fidelity and good rendering of fine details? and FIREPRO are better for complex models?
I've found that for my uses Quadro is better for fine details AND complex models.
My v7900 has been collecting dust since one week after I bought it… I really tried to love it, but ended up buying a Quadro 4000 instead. I just could not stand finetuning smooth curves when they were displayed jagged and flickered when I evaluated them… I nerded around, finetuned all settings, and the mesh and hatch antialias was smoooooth, but the curves were awful, and that mattered so much to me that I turned to the laptop to do the finetuning, while I was waiting for the quadro. It just felt like the Rhino viewports was running at the wrong resolution.
Yep, same as Holo here.
My Firepro has been relegated to (over)powering a home box and I’ve gone to a Quadro 4000.
The Firepros are great cards but as you and many others have found, the AA on curves in Rhino is just too horrible to work with for any period of time
turn Aliasing OFF, then use this command:
I am about to give you my project file to see how is your GPU workflow with this file. I need to see if it worth it
Your GPU QUADRO is for the desktop? or laptop?
I am thinking in change my GTX 780m (laptop) for the K3100M QUADRO the max version that may machine supports (but i did not found yet good arguments for it) … Sometimes GTX is better, because of many CUDAS i have. Other times it feel delays in viewports, specially When projects Became more and more complex, with lots of surfaces and polygons.
The other day I asked for a budget QUADRO K3100M at the store where I bought the laptop. I’m still waiting. Nevertheless, intent to invest in that Graphic Card (QUADRO K3100M) if i really going to get better performance visualization in softwares I use. Especially in Rhino … But also in RENDER, like V-Ray, or Artlanits, or OCTANE.
I know that the discussions around the graphics cards are endless and possibly the GPU FRAME only work well if only the recommended work by the same graphical software. But! my problem is that all software use OPENGL, and GTX are for GAMERS DIRECTX … In this perspective I am always in doubt whether or not I should invest in a professional graphics card
Does anyone have a friendly word that recommend?
Just curious, is the V7900 anti-alias in any other 3d programs bad?
Yes, but not as noticeably, as Rhino has superior curve AA.
Again, this is a line/curve thing as the rendermesh is smoothly AA’ed.
Comparing the relative speed of cards can be difficult. Many benchmark tests focus only on specific display or model types.
For instance many 3d benchmarks focus on shaded views of mesh models with or without textures. These benchmarks tend to favor the Gamer cards (GeForce and Radeon) card. This type of shading is used heavily in rendered viewport display and other Rhino display modes that do not display edges. Here is an example of this:
http://www.videocardbenchmark.net/high_end_gpus.html . Looking at this chart you would be crazy to buy anything but a GeForce card if performance was important. The $5000 K6000 shows up 7th in the list with 5 sub-$1000 cards ranked a head of it.
But, Rhino also needs to display many curves and edges which are OpenGl vector drawing. If benchmarks are focused more on this type of engineering display, the relative speed of cards can be quite different: http://www.tomshardware.com/charts/workstation-graphics-2013/20-OpenGL-SPECViewperf11-Solidworks-03,3296.html . Now the fastest GeForce does not show until card 25 of the fastest cards.
I would expect Rhino to benchmark out more like the engineering benchmarks. The HoloMark 2 benchmark may show this also.
This has been a pattern we have seen for years from the video card companies.
There is also the consideration of whether the GPU will be used for Cuda/OpenCL rendering.
See this list of comparisons of CUDA performance.
The top of the list is all the Supercomputer setups eg. 7x GTXTitans etc., but lower down the list are comparisons of single GPUs. Michael VS