Best GPU Graphics Card Options for Vectorworks in 2023 (and beyond)

[Asemblance]

Wondering if somebody from Vectoworks could give an update on the best GPU to get the most bang for buck in the software.

 

When buying new hardware for our studio, sellers try and push NVidia Quadro cards as they are supposedly more geared towards studio environments. However previous feedback from the Vectoroworks team and subsequent discussions suggest Quadro cards carry no particular benefit for Vwx, and you're best off going for the comparable GeForce card which will give best performance for the price.

 

Is this still the case? Any guidance on where vwx is at now in terms of GPUs - and where its headed in the future? Are there any advantages to the Quadro cards?

[_James]

I would also be interested in some more information from the VW team on this - @JuanP, do you have any more info on this or know who the best person to ask is?

[Jeff Prince]

I don’t work at Vectorworks, but I would base my purchasing decision on what my rendering and other softwares need instead of Vectorworks.

 

So, for Twinmotion, I recently chose the NVIDIA  RTX A4000 for our windows machine, more than sufficient for VWX, silky smooth for TM.

 

[_James]

@jeff prince Does Twinmotion take advantage of the extra features of A-series cards over the 30 or 40 series gaming focused cards? For a long time I have heard that VW doesn't and nor does Twinmotion.

 

Whilst I appreciate "bigger is better" generally on the GPUs, I am more interested in how will Vectorworks be using graphics cards moving forwards, will the VGM continue to expand? Will Renderworks go to the GPU? Hidden Line? Multi-GPU support? Will there be a point (or are we at it already) where the GPU is more important than the CPU?

 

An update to this article would be helpful as the hardware has come quite a long way in 4 years (as has VW) - 

 

[Jeff Prince]

4 hours ago, _James said:

Does Twinmotion take advantage of the extra features of A-series cards over the 30 or 40 series gaming focused cards? For a long time I have heard that VW doesn't and nor does Twinmotion.

 

I don't know.

I looked at the Twinmotion/Blender/ recommendation and chose the card that balanced price, performance, amount of VRAM, and power consumption.

 

[BartHays]

I may be out of my league here but when I was researching GPUs before the Nvidia RTX era the primary difference between Quadro Cards and GTX cads was the certification. There were some tweaks on Quadro cards to allocate more memory for handling complex geometry, whereas the GTX cards favored texture and shader processing. However, the biggest differences were that the Quadro cards were QC tested in-house to have higher levels of reliability and they used ECC memory chips to improve reliability. The GeForce cards were allowed to have some faults in memory and still be considered "in-spec." 

 

I am not sure if that is still the case, but it might be interesting to again, do side-by-side spec checks of Quadro and RTX cards to see if what you are paying (a lot)  for is that extra level of reliability.

 

Bart

[_James]

On 2/21/2023 at 9:41 AM, _James said:

Multi-GPU support?

Today Twinmotion announced that their next version will support multiple GPUs. Perhaps this will lead to offices buying computers with multiple GPUs and it would be good if VW could take advantage of that if there is a benefit in doing so.

[Jeff Prince]

3 hours ago, _James said:

Today Twinmotion announced that their next version will support multiple GPUs. Perhaps this will lead to offices buying computers with multiple GPUs and it would be good if VW could take advantage of that if there is a benefit in doing so.

 

Where did you hear that?

[_James]

 

The question is asked at 35:55 and their Senior Product Manager Colin Smith confirms that it will support it "for rendering" - perhaps that means just path tracer? They then go on to say they will have more information on how it works and how to set it up later which makes me think it may be limited in functionality and not be purely "plug and play"