CUDA 6.5 Is Doing Great For Blender

[phoronix]

Phoronix: CUDA 6.5 Is Doing Great For Blender

The NVIDIA CUDA Toolkit 6.5 update released this week provides measurable performance advantages to those using the open-source Blender modelling software with NVIDIA's GPU acceleration...

http://www.phoronix.com/vr.php?view=MTc3MDY

[liam]

Unless you're invested in cuda, or have to use cuda libraries, I don't understand why a project would use it. The only people who will ever(likely) benefit are Nvidia users.

[Daktyl198]

Unless you're invested in cuda, or have to use cuda libraries, I don't understand why a project would use it. The only people who will ever(likely) benefit are Nvidia users.

It probably just so happens that many Blender devs have Nvidia cards, and CUDA _does_ allow for some pretty awesome render times in Blender, so it was put in. As long as they don't try and maintain 15 different versions of CUDA (which it appears they are not), I'm fine with it since it probably doesn't take up too much of their time.

That being said, the sooner NVidia gives up and starts using OpenCL, the better ^.^

[liam]

It probably just so happens that many Blender devs have Nvidia cards, and CUDA _does_ allow for some pretty awesome render times in Blender, so it was put in. As long as they don't try and maintain 15 different versions of CUDA (which it appears they are not), I'm fine with it since it probably doesn't take up too much of their time.

That being said, the sooner NVidia gives up and starts using OpenCL, the better ^.^

I'm sure it is helpful for their nvidia users, and the person doing this work may not be familiar with, or even care about, opencl. Nvidia would be foolish to abandon cuda for openc since cuda gives them lockin and their fans don't seem to care about that.
As long as cuda at least stays even with opencl they will benefit from its continued development.

[magika]

Unless you're invested in cuda, or have to use cuda libraries, I don't understand why a project would use it. The only people who will ever(likely) benefit are Nvidia users.

Fans or not, when chosing between CUDA that works and does its job (e.g. yielding perfomance wins) or AMD OpenCL implementation that fails to run simple kernels people that care about their jobs would pick CUDA.
Its nice to tell fairy tales about open standards and such and bash proprietary things but when you want to do your job you simply choose proper tools.

[tuke81]

Afaik blender supports opencl too:

http://wiki.blender.org/index.php/OpenCL

And they want to support opencl and amd, but amd's drivers are not up to task:

http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/GPU_Rendering

What renders faster, NVidia or AMD, CUDA or OpenCL?
Currently NVidia with CUDA is rendering faster. There is no fundamental reason why this should be so?we don't use any CUDA-specific features?but the compiler appears to be more mature, and can better support big kernels. OpenCL support is still being worked on and has not been optimized as much, because we haven't had the full kernel working yet.

I'm afraid nvidia won't be so keen on opencl, they probably have prioritized for cuda then openACC next directcompute and maybe just maybe a tiny bit OpenCL.

[Aleve Sicofante]

OpenCL is a total disaster and almost everyone involved in doing compute tasks on graphics cards will choose CUDA over OpenCL.

It's easy to criticize Nvidia for going their own way, but when you see most projects are using CUDA instead of OpenCL for good reasons, you should at least ask yourself why. Radeon cards are usually cheaper and faster than Geforce ones. Firepro are MUCH cheaper than Quadro. But people keep buying Nvidia. And this is not casual buyers, but conscius, well informed ones.

While Blender has been trying to use OpenCL for a while (and will probably keep on trying), ask their developers. It's a nightmare and it's simply too hard to make it work. The fact is, until AMD gets its act together and puts more effort on the software side of their business (drivers and OpenCL), people will be better off with Nvidia.

[GraysonPeddie]

Hah... I would like to make use of my AMD A10-5700 APU's built-in GPU, but too bad. It would be nice if there's an nVidia GPU that does not have video outputs but are very affordable. It will just be a single-slot GPU which I could just turn it off when not in use. I'm not sure if nVidia binary blobs are compatible with Linux kernel 3.16, though.

[Marc Driftmeyer]

OpenCL is a total disaster and almost everyone involved in doing compute tasks on graphics cards will choose CUDA over OpenCL.

It's easy to criticize Nvidia for going their own way, but when you see most projects are using CUDA instead of OpenCL for good reasons, you should at least ask yourself why. Radeon cards are usually cheaper and faster than Geforce ones. Firepro are MUCH cheaper than Quadro. But people keep buying Nvidia. And this is not casual buyers, but conscius, well informed ones.

While Blender has been trying to use OpenCL for a while (and will probably keep on trying), ask their developers. It's a nightmare and it's simply too hard to make it work. The fact is, until AMD gets its act together and puts more effort on the software side of their business (drivers and OpenCL), people will be better off with Nvidia.

What a crock of crap.

[Calinou]

Fans or not, when chosing between CUDA that works and does its job (e.g. yielding perfomance wins) or AMD OpenCL implementation that fails to run simple kernels people that care about their jobs would pick CUDA.
Its nice to tell fairy tales about open standards and such and bash proprietary things but when you want to do your job you simply choose proper tools.

Because working on NVIDIA only and being biased is proper now?