GPU - hopping on the bandwagon / best bang for buck?

[Raph4]

I just read the thread and was wondering what IPR is?

Interactive Photorealistic Rendering

V-Ray production renderer in IPR mode WIP
https://www.youtube.com/watch?v=tK2eYCaMkI0

[dimension3plus]

Thanks for the answer and the video. This feature looks amazing.
I'm looking forward to it.

[savage309]

About the memory - if you have access to the nightlies they are already using far less memory for textures.
We have also the new mode allowing to load more textures than before as well (on-demand mip-mapped textures) in Max (in maya as well, but it is not yet tested at all there). You can test it with the *next* nightly (we fixed an bug there today) and give feedback (any will be helpful).

Best,
Blago.

[Libertyman86]

We need this benchmark comparison asap.

Another alternative that I'm contemplating is 2x1080s instead of the one Titan. That gives another 2000 cores (about 5,000 compared to 3,072 on the Titan) and better interactivity in the viewport with RT. It is however 'only' 8GB compared to 12GB. Price about the same.

I haven't seen any benchmarks comparing these alternatives yet though.

[Nicinus]

I have a friend in the cad industry that now have a system with two 1080s, founders edition. Unfortunately it is a work computer and I can't just install Max and Vray on it, are there any benchmarks that would be similar to using Vray RT that would be a bit more discreet that could give an indication ho fast it would be compared to only using one?

[mitviz]

over at the fstorm forum they have a benchmark going that shows you the different cards comparisons but again its using fstorm so idk how that works for you

[Nicinus]

Thanks, I'll take a look. I really just want to get a feel for how 2 1080s compare to 1 Titan X as I'm looking to get a new card and those represent similar investments.

[mitviz]

Thanks, I'll take a look. I really just want to get a feel for how 2 1080s compare to 1 Titan X as I'm looking to get a new card and those represent similar investments.

o they are faster! one 1080 was about 11 mins 15 secs and one titan x about 10mins out of the box, i have a zotac 1080 myself so i ran that test and my friend has the titan x so 2x1080 for sure beats up the titan x anyday

[werticus]

So the best solution would be two titan x's then :P

Any news of nvlink being viable? If two 1080's can share ram.... *wink wink*

[Nicinus]

So the best solution would be two titan x's then :P

Any news of nvlink being viable? If two 1080's can share ram.... *wink wink*

Well, if it worked it would be all the rave so I'm guessing not yet.

[savage309]

From what I know, NVLINK will be supported on some of the GP100 chips (that is Tesla P100). It will be not supported on GP102 (Titan X), GP104 (1080, 1070), etc. The memory (and actually the whole architecture) of GP100 is quite different.

Best,
Blago.

[werticus]

Oh, that means it needs the mezzanine connector, which is not a PCI-E card at all... So not much chance you'll be able to put this into an ordinary computer at all.

http://www.anandtech.com/show/10433/...ess-tesla-p100

[Preisler]

Does it matter which brand to look for when considering the 1080 card? ....or is it just best bang for the buck?

/Lars

[mitviz]

one with best cooling it seems but it really doesnt matter, people are buying the cheapest one they can also some advice stay away from the 1080 buy some 1070s