Page 7 of 10

Re: GeForce GTX 1080

Posted: Mon May 30, 2016 2:00 pm
by sigurdhoeyen
Okay,

So does anyone have any idea, when it i likely that we will see anything from Otoy regarding the new cards?

I'm just so dang curious...

Re: GeForce GTX 1080

Posted: Mon May 30, 2016 5:16 pm
by Gray_Parrot
Above, Glimpse links to another thread in which Octane says they have to test the new Pascal cards, which was dependent on the release of CUDA 8.0 from Nvidia. This information is confirmed in the Tom's Hardware review, which states:

"In the same vein, we had a lot of requests for Otoy’s OctaneBench. Just so you know, we tried to include results in today’s piece, but were stymied by a lack of support for GeForce GTX 1080 and 1070 in the demo version of OctaneRender. Nvidia acknowledges that it's sometimes necessary to work with developers before an application will recognize unreleased hardware, and in this case, Otoy will likely need to update its software to detect Pascal-based GPUs."

Re: GeForce GTX 1080

Posted: Tue May 31, 2016 6:33 am
by Elvissuperstar007
Arion Cuda test 1080 GTX = 3006 980Ti = 2288

Re: GeForce GTX 1080

Posted: Tue May 31, 2016 6:36 am
by Seekerfinder
Elvissuperstar007 wrote:Arion Cuda test 1080 GTX = 3006 980Ti = 2288
That's great news...

Re: GeForce GTX 1080

Posted: Tue May 31, 2016 9:31 am
by Synthercat
So they should give an OctaneBench score of 164 if it has the same performance increase in Octane too.

In that case the question is how many 1080 can you fit in your system and
1. Use only the pci-e 16x for fast times transferring models (2 ports in my case)
2. Use all available ports (8x 4x) for more cards but loosing some time transferring the model (I do a lot of animations and won't prefer that case)

Re: GeForce GTX 1080

Posted: Tue May 31, 2016 12:04 pm
by glimpse
Elvissuperstar007 wrote:Arion Cuda test 1080 GTX = 3006 980Ti = 2288
Thanks for sharing, Elvis! =)

Great piece of information (that again, correlates nicelly with SP numbers!)

Great news to say the least!

Re: GeForce GTX 1080

Posted: Tue May 31, 2016 12:44 pm
by sigurdhoeyen
That is really interesting to say the least, but my question still stands - when will we see the new architecture implemented?

Re: GeForce GTX 1080

Posted: Tue May 31, 2016 12:44 pm
by oguzbir
Hi Tom,
I've read that the 1080 will only support 2 way SLI.

Since we wont need SLI for Octane or other gpu rendering.
if so would it matter if we stack use 3 of them in the system.?

Apart from that . I currently have an old 680 + TITAN + TITAN.

I'll ditch the 680.
Set titan as primary display.
Sell the other Titan

And purchase 2 or 3 1080's
I think it'll be fine. Don't you think?

Thanks,

Re: GeForce GTX 1080

Posted: Tue May 31, 2016 1:01 pm
by glimpse
oguzbir wrote:Hi Tom,
I've read that the 1080 will only support 2 way SLI.

Since we wont need SLI for Octane or other gpu rendering.
if so would it matter if we stack use 3 of them in the system.?

Apart from that . I currently have an old 680 + TITAN + TITAN.

I'll ditch the 680.
Set titan as primary display.
Sell the other Titan

And purchase 2 or 3 1080's
I think it'll be fine. Don't you think?

Thanks,
ou, it's going to be fine, 'cos two way SLI will be supported "nativelly" by Nvidia itself, though..it will be possible to be enabled if needed (You just need to get special key..)

As we [OctaneRender users] are not using SLI it makes no difference for us at all =) so..I doubt we see any issues with multiple cards =) I would not worry about this at all.

one of thre reasons nvidia made this move is because 3/4way scalability is not perfectr & it's a lot of work + it's very little market for it. So they focus on 2way (with faster bridges) to support much better, giving possibility for developers (and end users) to go higher if needed..

Re: GeForce GTX 1080

Posted: Tue May 31, 2016 1:04 pm
by oguzbir
glimpse wrote: ou, it's going to be fine, 'cos two way SLI will be supported "nativelly" by Nvidia itself, though..it will be possible to be enabled if needed (You just need to get special key..)
As we [OctaneRender users] are not using SLI it makes no difference for us at all =) so..I doubt we see any issues with multiple cards =) I would not worry about this at all.
one of thre reasons nvidia made this move is because 3/4way scalability is not perfectr & it's a lot of work + it's very little market for it. So they focus on 2way (with faster bridges) to support much better, giving possibility for developers (and end users) to go higher if needed..
Cheers mate! Fantastic quick reply as always.