Ahmet,should we expect improved XP support in 2020.2?
(SDK XP made available OTOY)
Tell us what improvements can you expect now?
Thanks
Moderator: aoktar
VVG wrote:Ahmet,should we expect improved XP support in 2020.2?
(SDK XP made available OTOY)
Tell us what improvements can you expect now?
Thanks
Goldorak wrote: We are working to wrap up 2020.2 this month. We are on RC2 about to come out now, RC3 by end of week, and typically there are four RC builds before stable.
bepeg4d wrote:The issue is more complex, and it is caused by Nvidia CUDA.
The core devs have prepared 2020.1 SDK based on CUDA 11 with A100 Ampere GPUs, and everything was working perfectly.
Then Nvidia has presented Ampere RTX 30X0, but based on new CUDA 11.1, and the core devs had to rewritten everything with new 2020.2 SDK to perfectly support the new architecture.
So the issue is already fixed in latest Standalone 2020.2 Release Candidate.
But now, all plugins have to be rewritten with the new SDK, and this is taking time, but we are almost there, and 2020.2 has so many new cool features, that it worths to wait a little bit more, trust me!
Happy GPU Rendering,
ciao Beppe
Antoncromas wrote:bepeg4d wrote:The issue is more complex, and it is caused by Nvidia CUDA.
The core devs have prepared 2020.1 SDK based on CUDA 11 with A100 Ampere GPUs, and everything was working perfectly.
Then Nvidia has presented Ampere RTX 30X0, but based on new CUDA 11.1, and the core devs had to rewritten everything with new 2020.2 SDK to perfectly support the new architecture.
So the issue is already fixed in latest Standalone 2020.2 Release Candidate.
But now, all plugins have to be rewritten with the new SDK, and this is taking time, but we are almost there, and 2020.2 has so many new cool features, that it worths to wait a little bit more, trust me!
Happy GPU Rendering,
ciao Beppe
Your competitor's engine had denoise support for 3000 cards on day one. Just saying. . .
tomsvfx wrote:Antoncromas wrote:bepeg4d wrote:The issue is more complex, and it is caused by Nvidia CUDA.
The core devs have prepared 2020.1 SDK based on CUDA 11 with A100 Ampere GPUs, and everything was working perfectly.
Then Nvidia has presented Ampere RTX 30X0, but based on new CUDA 11.1, and the core devs had to rewritten everything with new 2020.2 SDK to perfectly support the new architecture.
So the issue is already fixed in latest Standalone 2020.2 Release Candidate.
But now, all plugins have to be rewritten with the new SDK, and this is taking time, but we are almost there, and 2020.2 has so many new cool features, that it worths to wait a little bit more, trust me!
Happy GPU Rendering,
ciao Beppe
Your competitor's engine had denoise support for 3000 cards on day one. Just saying. . .
Thats because Redshift uses Optix wich is made by Nvidia, so it makes sense that by default upgrading CUDA, Optix gets upgraded automaticly, probably without doing much.
Octane on the other hand has custom made Denoiser, wich also works way better than Optix (plus Optix has a big dedicated team working on it)
Your competitor's engine had denoise support for 3000 cards on day one. Just saying. . .
Thats because Redshift uses Optix wich is made by Nvidia, so it makes sense that by default upgrading CUDA, Optix gets upgraded automaticly, probably without doing much.
Octane on the other hand has custom made Denoiser, wich also works way better than Optix (plus Optix has a big dedicated team working on it)
I'm 90% sure that Octane has the same OptiX denoiser. And no, OptiX is not updating automaticly. Nvidia gives you CUDA SDK and OptiX SDK. And to implement it takes time. And i think you are right an OTOY team took some time when working on denoiser to make it custom. Because Redshift is bad at denoising. like alot.
Users browsing this forum: Baidu [Spider] and 5 guests