Forums » CentiLeo for Cinema 4D
Pages: 1
Does CentiLeo support NvLink?
 
LU
User  Posts: 17
Sep 2, 2021 12:07
Just like toptic, Does CentiLeo support NvLink? like if I make two RTX A5000 Cards, can I use each 24G Vram to make a 48G VRam in CentiLeo via Nvlink?

and I also read this on the website of PNY, It says that "The NVIDIA RTX A5000 Hardware accelerated Motion BVH (bounding volume hierarchy) improves motion blur rendering performance by up to 7x when compared to previous generation." so Does CentiLeo utilize this hardware to accelerate motion blur?
Edited: LU - Sep 2, 2021 12:18 (add something)
 
Administrator  Posts: 937
Sep 2, 2021 12:24
No, CentiLeo doesn't leverage Nvlink yet.
Nvlink when used makes visible 48GB, but the own onboard 24GB part works normally, the 24GB from another GPU are not as fast as from the own GPU.
CentiLeo has very efficient out-of-core tech for geometry and particles. You can populate the whole RAM with your geometry data and work with it on GPU, even if the GPU geometry cache is 4 or even 10x smaller than geometry storage in RAM.
CentiLeo Chat: https://t.me/centileochat
 
LU
User  Posts: 17
Sep 2, 2021 12:59
does that mean when I choice video card, the CUDA numbers is the most important part for rendering using CentiLeo? More CUDAs, faster speed?
 
Administrator  Posts: 937
Sep 2, 2021 13:31
Yes, btw any 3080, 2080, 1080 are perfect. 3090 is great!
Also for out-of-core a good PCI connection, RAM and CPU speed would make a benefit.
The first article about CentiLeo out-of-core sheds the light on what can help. The second article recommends the hardware for variuos purposes
https://centileo.com/docs/index.php?COURSE_ID=3&LESSON_ID=43&LESSON_PATH=17.43
https://centileo.com/docs/index.php?COURSE_ID=3&LESSON_ID=44&LESSON_PATH=17.44
Shortly the rule is: use at least PCI Express Gen3 X8 connection or better, RAM DDR4 as much as you need and powerfull enough CPU for you multi-GPU rig.
CentiLeo Chat: https://t.me/centileochat
Pages: 1
Users browsing this topic