The speed of Visualize is dertermined by the number of CUDA cores. The more CUDA cores, the faster Visualize will render. So the 1080s will be faster. Also, it's not recommended to run Visualize with SLI.
could you please elaborate on why SLI is not recomended. Are plans in place to support SLI in future?
you dont need to run your 2 980s in SLI to utilize all the cuda cores on both GTX cards.
SLI is used for improving realtime gaming performance. Currently, SOLIDWORKS CAD and Visualize use our own internal realtime engines, so SLI has no effect on realtime graphics.
SLI with raytracing (Iray) has no impact. So it's not so much that it's not recommended, just that SLI has no effect in Visualize, due to how we are using the NVIDIA GPUs for compute.
Years ago, using SLI and Bunkspeed sometimes produced unexpected results, so it's just better to disable SLI since it has not benefit anyways.
Hope this helps.
Thank you very much!
It seems I have to upgrade my graph card to GTX1080ti. So expensive !
or the other way, I build a new computer within one of my gtx980 and run the visualize boost on network.
Maybe the 980 + 980 on network will fast than 1080ti in one system,
What do you think ?
Yes, using Visualize Boost is for sure another option. You can put GPU budget into these Visualize Boost machines and use network rendering and the integrated Render Queue to crush your render jobs.
It's difficult to say since we only benchmark Visualize on NVIDIA Quadro cards. However you could start a new thread in the Visualize forum asking users on GTX performance benchmarks, and what they think you should do.
if you compare the 980 x2 vs 1080ti based on raw computing power the 980 making 4.9 tflops x2=9.8 tflops and the gtx1080= 11.5tflops
but if you are looking for best performance per dollar i suggest 1060 3g array 4 pcs and 4x4=16tflops
but pcie isnt always available in low cost system so 1080ti will be 2nd feasible option, but trust me im using 1070+1060 configuration its very fast in rendering you will love it,,,