What's better to accelerate the viewport: gtx 970 SLI or gtx 970 + gtx 1070 together?

A few months ago I bought one GTX 970 with the intention to get another one in the near future and SLI them together. Now that the GTX 1070 finally came out, I’m undecided about getting another GTX 970 to SLI them or get a GTX 1070 and make it work together with the GTX 970 thanks to the new DirectX features. I care more about accelerating the viewport than gaming cause I don’t play that much and when I do I play in 1080p and any of them are capable enough. So I’ll repeat the question one more time: what’s better to accelerate the viewport, two GTX 970’s in SLI or a single GTX 970 working together with a GTX 1070? Thanks for reading me!

OK, thanks, I’m still learning!

Hi, you cant accelerate the viewport with better or more cards.
You can accelerate render in viewport or final F12.
Render with two 970 is 200% of one card (or close to), the 1070 should be a bit faster but is not supported from Cycles render engine.

Cheers, mib

What I mean is be able to increase the level of subdivisions of my models. I did a multiresolution modifier test with Suzzane and at subdivision 7 (16,000,000 triangles) the viewport don’t lag. When I apply subdivision level 8 (64,000,000 triangles) the viewport becomes very laggy and I would like to improve the number of subdivisions that I’m able to apply because there are some bump maps that I would like to use as displacement maps for sculpting that don’t look good enough with the present number of subdivisions. I thought that a second video card would help me with that but it seems that I was wrong. What can be done to achieve it then?

There is something wrong with your workflow. You shouldn’t need more subdivisions than max 3, 2 is usually enough for everything. Perhaps you forgot to press smooth on your object? But anyway, you need to do something else to achieve what you are doing, than just leveling up subdivisions.