Importance of double precision performance?

A public forum for discussing and asking questions about the demo version of Octane Render.
Forum rules
For new users: this forum is moderated. Your first post will appear only after it has been reviewed by a moderator, so it will not show up immediately.
This is necessary to avoid this forum being flooded by spam.
Post Reply
User avatar (Default)
P!X3L
Posts: 19
Joined: Wed Jun 16, 2010 2:11 pm

Hello everyone!

I would like to know if the double precision performance of a gpu affects rendering times at all? I have heard time and time again that the new tesla cards will be about as fast as the fermi gtx cards. However, because the new tesla's have much more double precision performance; does that not debunk this "theory"? I have not seen any comparisons regarding rendering speeds between these two cards (because the c2050 is hard to get atm and the c2070 isn't out yet).

Any input would be appreciated. Thanks!

edit: BTW I'm getting a tesla c2070 as soon as I can get my hands on one. In an earlier post I mentioned that I will wait for opencl but that could take a while.... :D No worries, decided to go bonkers and go nuts with my next build. I may even get two of them if the price drops a few hundred euro after a few months of release.
Top
User avatar
radiance
Posts: 7633
Joined: Wed Oct 21, 2009 2:33 pm

Hi,

We're currently not using double precision as this would make octane incompatible with older GPU cards. (like 8000 series)
We might in the future, but only for a few small things, so it won't have any noticeable impact.

Radiance
Win 7 x64 & ubuntu | 2x GTX480 | Quad 2.66GHz | 8GB
Top
User avatar (Default)
GeorgoSK
Licensed Customer
Posts: 220
Joined: Sat Apr 24, 2010 4:01 pm
Location: Prague/Bratislava ;- )

What do you mean... ? The double precision is much faster on fermi based Tesla than it was on previous, but it's still roughly around half of the single precision performance. You can get flops measures on Nvidia's site on Wikipedia. There is also a 2050 review around here on forums that compares performance on it.

But they're still no way as powerful as regular GTX, they don't aim to, they need to run flawlessy for long time in supercomputers on so. So they're underclocked quite bit too.

That being said...I am jealous if you can afford one, since who woudn't want one ;) It's just bad you have to pay for all that stuff that will be literally useless just to get higher vram, which is only smaller part of the whole price addition compared to regular GTX.
Intel C4Q , 4G ram, GTX 285,
Top
Post Reply

Return to “Demo Version Questions & Discussion”