R15 Render Speed

Become a member of the CGSociety

Connect, Share, and Learn with our Large Growing CG Art Community. It's Free!

THREAD CLOSED
 
Thread Tools Search this Thread Display Modes
Old 07 July 2013   #16
Originally Posted by LucentDreams: I'm curious as to any cases in where the old physical methods would be faster than embree. I mean if there is an option what is the option there for? do the results differ notably, as they don't appear to, so are there cases where embree might strain compared to the old method, or limitations?


There are rare cases where the old method is faster than Embree. The main reason to keep the old method is its lower memory footprint. Scenes that push your RAM boundaries with the old method could run into problems with Embree.

There are three methods to choose from, Embree (Faster), Embree (Smaller) and Physical.
Physical has the lowest memory footprint, Embree (Smaller) up to 15-20% higher and Embree (Faster) up to 30% higher memory usage. The speed gain is well worth it though IMO.
__________________
http://www.bobtronic.com
 
Old 07 July 2013   #17
The only scene Ive tested where the new method is slower is a basic unrealistic scene with nothing but material displacement maps. Realistically once a scene is textured and lit, embree overtakes it again.
__________________
Matthew O'Neill
www.3dfluff.com
 
Old 07 July 2013   #18
Originally Posted by bobtronic: There are rare cases where the old method is faster than Embree. The main reason to keep the old method is its lower memory footprint. Scenes that push your RAM boundaries with the old method could run into problems with Embree.

There are three methods to choose from, Embree (Faster), Embree (Smaller) and Physical.
Physical has the lowest memory footprint, Embree (Smaller) up to 15-20% higher and Embree (Faster) up to 30% higher memory usage. The speed gain is well worth it though IMO.


Fantastic info Mathias, very good to know. I'll say I agree, seeing as how I have 10GB per physical core I can't imagine Embree fast being an issue. Good to know on really memory intensive scenes there's an option that can see a notable memory savings

Does embree work on AMD systems?
__________________
Quote: "Until you do what you believe in, how do you know whether you believe in it or not?" -Leo Tolstoy
Kai Pedersen
 
Old 07 July 2013   #19
Originally Posted by LucentDreams: Fantastic info Mathias, very good to know. I'll say I agree, seeing as how I have 10GB per physical core I can't imagine Embree fast being an issue. Good to know on really memory intensive scenes there's an option that can see a notable memory savings

Does embree work on AMD systems?


It works on any system supporting SSE3, no matter if Intel or AMD (as there isn't or hasn't been any switch code in Cinema prefering any of the vendors).

Best regards,

wbj
 
Old 07 July 2013   #20
Originally Posted by wbj: It works on any system supporting SSE3, no matter if Intel or AMD (as there isn't or hasn't been any switch code in Cinema prefering any of the vendors).

Best regards,

wbj


Okay, I was just wondering as embree is an intel technology if it was specially optimized with functions in their processors.
__________________
Quote: "Until you do what you believe in, how do you know whether you believe in it or not?" -Leo Tolstoy
Kai Pedersen
 
Old 07 July 2013   #21
Originally Posted by LucentDreams: Okay, I was just wondering as embree is an intel technology if it was specially optimized with functions in their processors.


You could of course configure the source code in a way that it would only run on SSE 4.2 or AVX only, but who (besides of the hardware vendor) would be interested in narrowing the potential customer base down to a fraction for getting 3-5% more speed?

Best regards,

wbj
 
Old 07 July 2013   #22
My colleagues and i tested at several amd cpus, at recent new public too.
Used corona standalone and several compiled cg apps. At most of amd it does not work.

small cross-post and rip from http://embree.github.io
Quote: Embree 2.0 now also boasts support for the Intel Xeon Phi coprocessor platform; support for high-performance "packet" kernels on SSE, AVX, and Xeon Phi; integration with both traditional C++ and with the Intel SPMD Program Compiler (ISPC); instantiation and fast BVH reconstruction; and an example photo-realistic rendering engine for both C++ and ISPC.

Last edited by ilay : 07 July 2013 at 08:08 AM.
 
Old 07 July 2013   #23
Is there any image quality differences? These examples don't show what the renders from R14 look like, only the R15...or are they identical?
__________________
MikeFix.com | C4DList.com
 
Old 07 July 2013   #24
Originally Posted by MikeFix: Is there any image quality differences? These examples don't show what the renders from R14 look like, only the R15...or are they identical?


Theyre identical
__________________
Matthew O'Neill
www.3dfluff.com
 
Old 07 July 2013   #25
if you pixel-peep you can find some very, very minor differences. nothing worth mentioning though.
__________________
http://www.bobtronic.com
 
Old 07 July 2013   #26
mainly like sampling grain differences?
__________________
Quote: "Until you do what you believe in, how do you know whether you believe in it or not?" -Leo Tolstoy
Kai Pedersen
 
Old 07 July 2013   #27
Originally Posted by LucentDreams: mainly like sampling grain differences?
Not really, there are some subtle differences in shadows and highlights, not in all cases though. But nothing to worry about as I find it now hard to find an example
__________________
http://www.bobtronic.com
 
Old 07 July 2013   #28
Interesting I wouldn't have expected lighting changes. I was thinking something like grain dithering or AA type behavior. Sounds like to be safe you choose one method on a project and stick with it.
__________________
Quote: "Until you do what you believe in, how do you know whether you believe in it or not?" -Leo Tolstoy
Kai Pedersen
 
Old 07 July 2013   #29
Originally Posted by LucentDreams: Interesting I wouldn't have expected lighting changes. I was thinking something like grain dithering or AA type behavior. Sounds like to be safe you choose one method on a project and stick with it.


As I said, it's barely visible. I only noticed it on some clay-render tests and I had to squint.
__________________
http://www.bobtronic.com
 
Old 07 July 2013   #30
Great Results

Is Sketch&Toon faster too?
__________________
64Bit - C4D R15 Studio, ZBrush 4R7, 3D-Coat, World Machine, Photoshop CS6 ext

...living in Paradise
 
Thread Closed share thread



Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
CGSociety
Society of Digital Artists
www.cgsociety.org

Powered by vBulletin
Copyright 2000 - 2006,
Jelsoft Enterprises Ltd.
Minimize Ads
Forum Jump
Miscellaneous

All times are GMT. The time now is 01:49 AM.


Powered by vBulletin
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.