(Hardware Testing - Graphics Cards)
Now we reach the crux of the matter. What makes a difference in Graphics performance with SolidWorks? We discussed in our first graphics article (FSWPC-13-#7a) that while some of the graphics related functions were fairly inert other made a significant impact, either positive or negative, but these effects were fairly consistent across the spectrum of graphics card options. So still we can't tell much of a difference between a Quadro 600 and a Quadro k5000; that is until we start diving a little deeper.
In our testing we ran three models through our functional test.
The large assembly that we used for all of our testing.
A small or medium sized assembly with roughly 700 components.
A single imported complex part.
We ran each model set through the exact same benchmark 20 times. Each run isolated a different functional option (ie RealView on or off, display state options, etc…). We then combined the results of each test into a "total runtime" and began comparing results relating different model sets and different cards.
The results were somewhat counter intuitive.
Our original assumption was that the bigger the model set the more the graphics card would be taxed. Thus, the more powerful the graphics card the faster it would be.
The most interesting conclusion that we have come to is that you cannot expect an increase in graphics performance when working with very large assemblies if you buy a more expensive graphics card.
The maximum percentage improvement found with large assemblies was 1.6% between the k5000 and the others which were practically identical. We found this to be very disheartening but continuing the tests showed the positive side.
Small assemblies and parts showed a much greater tendency to improve performance when increasing graphics card power and capability.
The Small Assembly and Part Model showed that improving the graphics card capability can greatly affect the number of frames a card can paint in a given time thus improving the user experience and comfort. Our tests ran as much as 52% faster with the k5000 rather than the Quadro 600. Knowing what we know about graphics performance this potentially means far smoother graphics transitions resulting in a better user experience.
There is a threshold however. If the graphics card is already pushing out 60 frames per second; will you be able to perceive an increase to 100 frames per second? There comes a point that it does matter if the card is going faster if you don't get to experience the change due to your own natural limitations.
So it comes to a recommendation. What card will work for your needs? What we have found through our testing and through our many years of SolidWorks experience is that the NVIDIA Quadro 2000 is the best card for SolidWorks users. Realizing that going with a higher level card will not help improve the performance of Large Assemblies and the Quadro 2000 allows us to experience the same level of performance as the higher level cards do to the limitations of the human eye.
It is very important to realize that when discussing performance of graphics cards speed can be a misleading term. Our testing measured the cards' speed in painting a certain number of frames in a simulation of spinning our test models from one location to another. The faster it paints the frames the faster it reaches the final position of rotation and finishes the test. In everyday modeling life the final rotation position would be reached at the same time by every card. The performance difference would manifest itself in slower tests painting fewer frames in the given time and thus producing a "choppier" visualization of the spin.
Also, SolidWorks has some built in tools to work around slower performing cards to still allow the user to experience a smooth spin. This usually involves removal of visualization detail to allow the card to paint faster than would be possible if all of the detail remained in the model.
The last item to consider when choosing a graphics card is to think about the other programs that you are running on your machine and if they can take advantage of the GPU that a higher power graphics card can offer.
Please check back to the CATI blog as we will continue posting our series of articles that goes further into the details of each of our tests. All of these articles will be stored in the category of Free SolidWorks from Performance Constraints and links to each with their release date are listed below:
- FSWPC-13-#1 Introduction (3/19/13)
- FSWPC-13-#2 RAM (3/21/13)
- FSWPC-13-#3 Processor Cores (3/26/13)
- FSWPC-13-#4 Processor Cores for Simulation and PhotoView360 (3/28/13)
- FSWPC-13-#5 Processor Clock Speed (4/2/13)
- FSWPC-13-#6 Hard Drives (4/4/13)
- FSWPC-13-#7a Graphics Cards-SolidWorks Functionality (4/9/13)
- FSWPC-13-#7b Graphics Cards-Hardware Comparison (4/11/13)
- FSWPC-13-#7c Graphics Cards-Model Complexity (4/16/13)
- FSWPC-13-#8 Files stored in an older version (4/18/13)
- FSWPC-13-#9 OS Visual and User Settings (4/23/13)
- FSWPC-13-#10 SolidWorks Options (4/25/13)
- FSWPC-13-#11 SolidWorks Image Quality (4/30/13)
- FSWPC-13-#12 SolidWorks RealView Graphics (5/2/13)
- FSWPC-13-#13 SolidWorks Level of Detail (5/7/13)
- FSWPC-13-#14 SolidWorks Mates and Flexible Subassemblies (5/9/13)
- FSWPC-13-#15 SolidWorks Defeature (5/14/13)
- FSWPC-13-#16 SolidWorks Envelope (5/16/13)
- FSWPC-13-#17 SolidWorks Large Design Review (5/21/13)
- FSWPC-13-#18 SolidWorks SpeedPak (5/23/13)
- FSWPC-13-#19 SolidWorks Save as Part (5/28/13)
- FSWPC-13-#20 SolidWorks Large Assembly Mode (5/30/13)
- FSWPC-13-#21 Optimal Configuration and Conclusion (6/4/13)
Josh Altergott, CATI Support Manager
Adrian Fanjoy, CATI Technical Services Director