Page 73 of 161

Re: NEW: Octane X - is finally here! [MEGA FAQ THREAD]

PostPosted: Tue Apr 27, 2021 10:31 am
by m0dm0d
Here's one more (if it's useful)

Vega56 = 26 seconds (egpu)
Image

Re: NEW: Octane X - is finally here! [MEGA FAQ THREAD]

PostPosted: Tue Apr 27, 2021 1:25 pm
by pegot
TheBloke wrote:One final question: in your result graph there is a NVidia 2070 Super, which I assume must be in Windows on CUDA. Are you able to get a NVidia 3080 CUDA test for the graph? I would love to see how the 6900XT on Metal compares to 3080 on CUDA.


So I just ran a test using a RTX 3080 and got 8 seconds (same time regardless of RT acceleration being turned on or not). Using Network rendering with an additional 3 GPUs (2 1080 ti's and 1 1070 ti) halved that to 4 seconds.
Rtx-3080.png

Re: NEW: Octane X - is finally here! [MEGA FAQ THREAD]

PostPosted: Tue Apr 27, 2021 5:35 pm
by stevelong03
Hello! Otoy. Can I ask about when the production ready version of Octane X will be release? Or at least use the newer SDK? 2020.2 or 2021? Because I also use the windows machine. The windows version is 2020.2.3. It’s great! So wanna to know about The Mac version. Thanks a lot!

Re: NEW: Octane X - is finally here! [MEGA FAQ THREAD]

PostPosted: Wed Apr 28, 2021 6:43 am
by bepeg4d
@m0dm0d: big thanks, any AMD GPU not listed is more than welcome :D

@pegot: big thanks to you too, now we have a reference from all the most recent Nvidia GPU families, Pascal Turing, and Ampere :D

Here is the updated graph:
288E4D32-CB5C-4B7F-AC73-F07DF2E00670.jpeg


@stevelong03: sorry I cannot share any ETA.

Happy Mac GPU rendering,
ciao Beppe

Re: NEW: Octane X - is finally here! [MEGA FAQ THREAD]

PostPosted: Wed Apr 28, 2021 10:40 am
by VjekoA
Here is my results. AMD RX 5700 + AMD RX Vage 64, together and single results.
RX5700 (internal GPU) with RX Vega 64 (internal GPU) = 11 seconds
RX5700 = 23 seconds
RX Vega 64 = 22 seconds
https://imgur.com/a/khgDnUI

Cheers!

Re: NEW: Octane X - is finally here! [MEGA FAQ THREAD]

PostPosted: Wed Apr 28, 2021 11:20 am
by tffx
Any idea when we'll be able to use Octane in C4D on an M1 Mac natively without having to open C4d under rosetta ?

Re: NEW: Octane X - is finally here! [MEGA FAQ THREAD]

PostPosted: Wed Apr 28, 2021 2:08 pm
by filipw
Great work. have a 6900xt in egpu that matches results mentioned here. My Linux PC with 3090 finishes in 7 secs. But, I have tested on more advanced scenes and in some scenes 3090 and 6900 is exactly on par while in some the 3090 pulls ahead a lot. All in all it feels like Octane is finally at a stage where it renders most scenes well on Mac.
I only have a difference in amout of fireflies when rendering on PR11 compared to linux 2020_2_1. Is there some change that has not been implmented on the mac yet?

Re: NEW: Octane X - is finally here! [MEGA FAQ THREAD]

PostPosted: Thu Apr 29, 2021 7:22 am
by bepeg4d
filipw wrote:Great work. have a 6900xt in egpu that matches results mentioned here. My Linux PC with 3090 finishes in 7 secs. But, I have tested on more advanced scenes and in some scenes 3090 and 6900 is exactly on par while in some the 3090 pulls ahead a lot. All in all it feels like Octane is finally at a stage where it renders most scenes well on Mac.
I only have a difference in amout of fireflies when rendering on PR11 compared to linux 2020_2_1. Is there some change that has not been implmented on the mac yet?


Great, thanks for the positive feedback :D

About fireflies, weird, it should not :roll:
Could you share with me, even in Private Message, a scene that shows the issue?

ciao Beppe

Re: NEW: Octane X - is finally here! [MEGA FAQ THREAD]

PostPosted: Mon May 03, 2021 4:47 am
by dladopoulos
Chess set benchmarks on a Mac Pro:

1 x AMD Vega II Duo : 18 sec
2 x AMD Vega II Duo : 9 sec
3 x AMD Vega II Duo : 6 sec
4 x AMD Vega II Duo : 4 sec

1 x AMD Vega II Duo.jpg

2 x AMD.jpg

3 x AMD.jpg

4 x AMD Vega II Duo.jpg

Re: NEW: Octane X - is finally here! [MEGA FAQ THREAD]

PostPosted: Mon May 03, 2021 9:15 am
by bepeg4d
Hi dladopoulos,
thanks :D
Do you mind to redo the test with 2x GPUs?
It should be 8/9 :roll:

ciao Beppe