SOLIDWORKS 2020 SP1 GPU PerformanceWritten on February 28, 2020 by William George
With the recent release of SOLIDWORKS 2020's first service pack, we thought it would be a good time to do a roundup of NVIDIA Quadro video card performance with regards to modeling in SOLIDWORKS. We are leaving GeForce out of the running this time, because they are not officially supported by Dassault Systèmes and some graphics options cannot be enabled without a supported card.
Looking for an Engineering Workstation?
Puget Systems offers a range of powerful and reliable systems that are tailor-made for your unique workflow.
Here are the detailed specs of the test platform we used, along with a list of the Quadro cards we included:
|CPU||Intel Core i9 9900K|
|CPU Cooler||Noctua NH-U12S|
|Motherboard||Gigabyte Z390 Designare|
|RAM||4x DDR4-2666 16GB (64GB total)|
|Video Card||NVIDIA Quadro RTX 6000 24GB
NVIDIA Quadro RTX 5000 16GB
NVIDIA Quadro RTX 4000 8GB
NVIDIA Quadro P6000 24GB
NVIDIA Quadro P2200 5GB
NVIDIA Quadro P1000 4GB
|Hard Drive||Samsung 960 Pro 1TB|
|Software||Windows 10 Pro 64-bit
SOLIDWORKS 2020 SP1
This platform is built around an Intel Core i9 9900K, with very high clock speeds, to avoid the CPU being a bottleneck in this testing. That processor also gives great performance in SOLIDWORKS for general usage and modeling. More than enough RAM was included, to avoid that being a bottleneck of any kind, and a super-fast M.2 SSD was used for the same reason. For the video cards, we included most of the Quadro RTX line - except for the RTX 8000, which should perform the same as the RTX 6000 but with more VRAM - as well as the older Quadro P6000 and some of the lower-cost P-series models.
To perform the actual benchmarking, I used the same basic testing we've used here at Puget for analyzing graphics performance in SOLIDWORKS in the past, just updated slightly for the 2020 release: a mix of AutoIt scripts and SOLIDWORKS macros to set the different quality settings, load the relevant model, and record the average frames per second while rotating the model. To determine the frame rate, a macro is used with a timer to rotate the model 45 degrees to the left and right for a set number of frames. From the number of frames and the total time it took to render those frames, our software is able to determine the average FPS.
For test samples, we have utilized models available from GrabCad.com that provide a range of complexities based on the total number of parts and number of triangles - along with one extremely large assembly provided by Daniel Herzberg, who has organized the CAD Monkey Dinner at SWW / 3DExperience World for the last several years. This time, however, I focused on just two assemblies:
I decided to omit results from the smaller assemblies in our test suite, because they are so fast that the data would have been repetitive. By focusing on a mid-size project and a very large one we can better see how these video cards behave. If you work exclusively with smaller assemblies or parts, and don't have any other demanding 3D graphics needs, then any of these Quadro cards should be fine.
Results & Analysis
Here is a gallery showing SOLIDWORKS performance from each of the tested video cards with the Audi car assembly, at both 1080P and 4K resolutions, and with a variety of quality settings:
The framerates recorded with this assembly are so high as to be absurd, which is part of why I decided to omit the smaller projects from this article. All of the tested cards did just fine at 1080P resolution, though the P1000 got into potentially-uncomfortable territory at 4K.
And here are the results from the Lego Tower model, again at both 1080P and 4K resolutions and with the same quality settings:
The story is quite a bit different with the more complex Lego Tower assembly, which visbly crawled with the P1000 and wasn't smooth on the P2200 either. Interestingly, the newer RTX 4000 outpaced the older P6000 - even though it costs only a fraction of the price. We still use a couple P6000s here in Labs when we need to test software that isn't certified on GeForce cards, but given how much they cost and how the whole RTX series outperforms them I don't think it makes sense to get the higher P-series cards for SOLIDWORKS today.
And finally, we tested the Lego Tower assembly with LAM (Large Assembly Mode) on as well - it was off in the previous set of results:
In past testing, I have usually turned Large Assembly Mode off for GPU comparisons - but this time around I wanted to see how it impacted performance, so I ran the Lego Tower (the only assembly we have which is large enough to trigger use of LAM) again. As far as I can tell, it doesn't seem to have improved the "normal" (Realview off) performance at all, but it did bring the Realview-enabled frame rates up to the same level. Given how close those results ended up being, I thought I should look up and refresh my memory on what exactly LAM does - and it turns out that it disables Realview. So I guess that test wasn't really very helpful! LAM is listed as doing a lot of other things too, but in terms of actual performance impact it is the same as simply turning off Realview... which it would do anyway.
Taking all of the results above into account, I have two general recommendations for video cards in SOLIDWORKS 2020:
- For mid-size and smaller assemblies, the Quadro P2200 is a great choice - and not very expensive (as workstation cards go)
- For larger, more complex assemblies, the newer Quadro RTX 4000 is a solid option
Only extremely massive assemblies, at the highest resolution and quality settings, would need a more powerful GPU. We do offer the whole line of Quadro cards in our engineering workstations, though, so you can select the one that best fits your needs and budget!