3ds Max Viewport performance 7980xe Vs 2699 V4

[karolismacernis1]

Hello,

 

First of all, apologies if the post is a bit long.

 

We are a looking to invest in new hardware for rendering at our architecture design studio for up to 4000 euros. We are using 3ds Max (Vray) on a 6900k machine with a gtx 1060 at the moment.

 

Our main goal is to increase rendering speed (for faster previews and final stills) and if possible viewport performance.

 

The main options that we are looking at is to upgrade the workstation with a 7980xe 18 core CPU or dual 2699 V4 OEMs. The scenes we work on range from a small bathroom to large projects in urban cityscape environments with many polygons. And since in every project a lot of time is spent generating fast low res previews on the local workstation I think its good to invest in having a powerful cpu onboard the main machine to save time.

 

I know that dual 2699 V4s would be about 30% faster than a 7980xe in rendering, however since doing some research, I found that a lot of people are recommending i7/i9 over Xeon for viewport performance as 3ds Max FPS while navigating the scene is dependent on the CPU single thread speed a lot as well, not only the video card. Does someone have experience in comparing these kind of systems? We are not considering AMDs threadripper that much because even though it is a great price/performance value, the lower single threaded performance might be noticeable in the viewport and even though 7980xe has about 30% performance increase in rendering, to my mind it is completely worth it, if it is used to generate revenue every day and since we are a small team we want to save as much time as possible in order not to turn down or delay projects.

 

Or do you think it is not worth to upgrade from a 6900k and instead invest in more render nodes and use them to generate previews with distributed rendering? Perhaps someone has some intake of how long it takes for DR to kick in when working on a scene and if it is necessary to invest in SSDs for that on each node?

 

Also would you think that upgrading to a better graphics card from gtx 1060 could lead to better viewport performance or it is not worth it? (We are not doing animations, only still rendering)

 

Aside from upgrading the main workstation we want to invest in a small render farm to push the work in order to continue using the main pc for modelling or use DR. We looked at several options for the nodes which are:

 

- a couple of AMD 1700s with a B350 motherboard (seems like a very cost efficient option. It is possible to build one these nodes for about 700€ which would have the same a computing power of a 6900k)

 

- buying some used xeons like the 2670 v1 and building a node with that or some used x5650. It is possible to buy one server with a dual x5650 for arround 200€ which gives you 1250 pts in Cinenbench. However, then you have a risk if all the used components will work properly. Anyone has some comments on this?

 

- building a single node with a 1950x threadripper (again AMD in this case because it gives a good value for rendering performance and single thread speed is not that important for a node.

 

Again, sorry if the post is a bit long. Opinions are appreciated.

 

Thanks a lot!