12GB VRAM for $3,000

[savage309]

Same goes for dual GPU cards unfortunately (as savage309 points out), but this may be changing in both cases soon from what I gather.

With current technologies it will be huge speed penalty to do so.
Faster ones (like nvlink) are coming in 2016

[abowen]

Same goes for dual GPU cards unfortunately (as savage309 points out), but this may be changing in both cases soon from what I gather.

so . please confirm for me ... this card will only be able to load 6gb of a proj into card , as it will divide the 12gb by the 2 processors? if that is the case the only benefit is form factor benefit? wow! harder sell

[Dariusz Makowski (Dadal)]

so . please confirm for me ... this card will only be able to load 6gb of a proj into card , as it will divide the 12gb by the 2 processors? if that is the case the only benefit is form factor benefit? wow! harder sell

Yea that's correct

You basically gain compactness. So you can stick more of them in ur workstation. Basically 4 pcie slots can either give you 4GPU or 8 GPU.

[savage309]

so . please confirm for me ... this card will only be able to load 6gb of a proj into card , as it will divide the 12gb by the 2 processors? if that is the case the only benefit is form factor benefit? wow! harder sell

That's correct.

edit : late response.

[vlado]

Yes! Video please!

Here is the recording of my presentation: http://nvidia.fullviewmedia.com/gtc2014/S4779.html

Though the video that I show with the nVidia cluster is really just playing around, we should do a more impressive one...

Best regards,
Vlado

[Dariusz Makowski (Dadal)]

Here is the recording of my presentation: http://nvidia.fullviewmedia.com/gtc2014/S4779.html

Though the video that I show with the nVidia cluster is really just playing around, we should do a more impressive one...

Best regards,
Vlado

Thanks Vlado! Very interesting stuff.

I quite like the cluster implementation, specially the option to dynamically add DBR nodes. Do you think that thing can be ported in any way to Maya ? or be cmd controlled?

[Pixelcon]

daaaaaaaaaaaaaaaaaaamn thats fast!

[vlado]

Do you think that thing can be ported in any way to Maya?

Yes, the 3.0 SDK does allow render nodes to be added (or removed) on the fly, we just need to figure out a good UI for this.

Best regards,
Vlado

[Dariusz Makowski (Dadal)]

Yes, the 3.0 SDK does allow render nodes to be added (or removed) on the fly, we just need to figure out a good UI for this.

Best regards,
Vlado

Wow thats awesome. It would be great if it can be standalone ui tho... then again I think it might be hard given different jobs, deadline and so on huh... still nice to know its coming

[FSGFX]

id like to run the vray cloud on this oil cooled nv cluster...

http://blogs.nvidia.com/wp-content/u.../kfc-nodes.jpg

[lukx]

so now just wait a year and get it for 1500$

[tricky]

Its always a damn car though. Cars are easy peasy

[Morne]

Its always a damn car though. Cars are easy peasy

+1
Yes I'm yet to get RT to work with my type of scenes

[eyepiz]

Its always a damn car though. Cars are easy peasy

+1
Every time I try RT GPU on an arch interior something is always not supported or broken.

[vlado]

Every time I try RT GPU on an arch interior something is always not supported or broken.

If you want to use the GPU successfully, it would be best to do your scenes with it from the start. Just taking an existing scene for the regular V-Ray and throwing it at the GPU will likely always come across some little option somewhere that's not supported or works slightly differently.

Best regards,
Vlado