Maxwell Render 4 to offer GPU rendering
Aug 17, 2016 by CGP Staff
3
|
Next Limit has announced the upcoming release of v4 of its renderer. New features include:
- GPU rendering – the new engine runs on NVidia graphic cards. All the technology under the hood is identical to the classic CPU engine – which means your images are exactly the same.
- Maxwell Multilight Standalone – save infinite lighting variations from one single render
- Lighting setups in real time
- Work faster by editing lights before and after the render finishes
- Create a lighting catalog for your client
- Automatically update your catalog with different lighting positions or geometry
- New materials gallery – the renovated gallery features a hand-picked selection of high-quality, optimized Maxwell materials ready to use in your scenes. You can search using the usual tags, such as wood, plastic or metal, but also by performance level – so that you can filter materials depending on your exact needs.
- Rhino for Mac support – new Rhino for Mac plugin workflow, providing a live-link between Rhino and Maxwell Studio
- Significantly improved Revit integration
Looks like Nextlimit ate crow with this one. I remember in the times of the GPU advent the devs said it makes no sense to go GPUs and that Intel would soon release a big cpu which years after ended up as Xeon Phi product.
Well that’s what you get for trying to be a smartass who knows how hardware will progress.
Still GPU’s will be very limited by vram.
And it seems that they have just recompiled their existing code for GPU so I think it is not as optimized as code made for just GPU.
Maxwell is great renderer. CPU still has advantage, it is not limited and in CPU you can use even the oldest processors for rendering when in GPU you must have specific hardware and drivers. CUDA and OpenCL and soon Vulkan based renderers. It is hard to mix Nvidia and AMD GPU’s for rendering but it is easy to use AMD and Intel CPU’s for rendering if x86 based architecture.
Xeon Phi is quite interesting idea, co-processors has potential. GPU has also potential but it must solve their vram memory problem (at the moment render engines can use only one gpu card memory even when you have 10 gpu cards) and I guess it will need totally new architecture design before all memory available can be used.
But I must admit that AMD Radeon Pro SSG graphics 1TB card is promising, we just need render engine supporting for it, even when it is using solid state storage it should be still faster than CPU and that memory would be quite good even for large and complex scenes with huge dynamic simulations. Maybe Arnold will release in near future GPU acceleration for this card because they have made co-operation with AMD and Solid Angle had presentation from GPU technology that it is on its way for Arnold Render.
Maybe for Maxwell, but not the others. There’s “Out of Core” now in Octane etc. So the 6gb is only a limit for the actual geometry and – with Out of Core – you can use your RAM memory for the textures. I haven’t hit the 6gb VRAM limit yet and will soon get 2x Titan X-cards. So memory isn’t a problem for most users.