Hot answers tagged

125

Ensure GPU Support Currently cycles supports CUDA (Nvidia) devices and has experimental support for OpenCL devices as of 2.75 (added in B7f447). If you are using an AMD/ATI graphics card, see the OpenCL section below. If you are running an older NVIDIA card, ie the Geforce series, support is extremely limited and these are not officially supported, see How ...


27

Update as of July 29th 2019: NVIDIA RTX GPUs are now officially supported and can be used to accelerate renderings in Cycles using the raytracing cores of RTX cards. You can read more about this from the Blender Developers blog here. In the past, RTX GPUs were not supported even for normal use cases unless you compile Blender from source code yourself or ...


22

I found one more setting to check that sets Blender to use CPU. I rendered using this script(gpurender.py): import bpy bpy.context.scene.cycles.device = 'GPU' bpy.ops.render.render(True) Command-line was: blender -b file.blend -E CYCLES -t 0 -o `pwd`/outpre -P gpurender.py If you check in python, bpy.context.scene.cycles.device will probably be CPU. I've ...


21

If you render an animation, you might achieve what you want by having two instances of Blender running. One renders with CPU, the other with GPU. Both use the same file and the same folder to render the image sequence. You can tell blender in the Output panel to create placeholders and to not overwrite existing files. Placeholder creates an empty file when ...


20

The difference and why there are different optimal bucket sizes comes from the size and design of on chip cache/memory of cpu or gpu. The gpu has a massive amount of cores. But they are dumb and cannot do much logic. They have a big data throughput though - so that's why is good to give them a big chunk of data. There will be a lot of parallelization done ...


18

I'm not sure if you really have any problem here. Actually here is common mistake made by a lot user that are not into programming. GPU rendering is NOT faster than a CPU rendering, GPU rendering is SOMETIMES faster than CPU rendering, depending on the rendering algorithm it's used and your graphic card your OS the pipeline etc... Blender don't use the ...


16

GPU usage is highly dependent on the scene complexity. With a very simple scene like the one you tested, it is hard to bog-down a modern GPU. But with a realistically complete scene, you can expect the GPU usage to be a lot higher. Take a look at this Cycles rendering benchmark


13

There are lots of good articles out there on this topic, and short of recreating them, I can say that the two cards I use are not in SLI (wrong motherboard for it) and it uses them like two separate cores. It'll render two chunks at a time, if you tell it to use both. Some of the articles say that running in SLI is a little slower than running two separate ...


13

Go to "File"->"User Preferences" and click on the "System" tab. On the left it should say "Cycles Compute Device". On my machine (I have the same card) when I change from "None" to "CUDA" the GPU Compute rendering option stops being greyed out. If that's not your case I don't really know what the problem could be. Good luck!


13

Blender Cycles relies on GPU compute languages like CUDA or more recently OpenCL. CUDA is exclusive to Nvidia and while OpenCL is technically cross platform, in practice AMD's implementation is the only one that provides a usable speedup for Cycles. From the Blender wiki: https://wiki.blender.org/index.php/OpenCL ... Intel GPU’s are getting more powerful ...


12

The SSS shader has not been implemented for GPU rendering yet. It will always render black. Update: SSS now works on GPU, so they should look the same. If you're looking for quick renders, SSS is not a good idea. If you can get away with a simple Diffuse BSDF shader with a bit of Translucent BSDF, that'd be great.


12

It is possible to render a movie on a single computer. It depends on what movie it is. If you choose a style that renders extremely fast, it is possible. Nothing like the ones Disney make though. Not even remotely close. Let's say you are rendering a movie that is 1 hour and 30 minutes. Let's say 1 frame of that movie renders extremely fast - 2 minutes. ...


11

Here's a spreadsheet that collates data from many users running the same benchmark scene. The thread with active discussion and a copy of the scene is at BlenderArtists, This is the best we can do without relying on opinion. Decide for yourself which gfx card makes sense. Note: Different OS's, drivers and versions of Blender, can all significantly change ...


11

You can, as long they are from the same brand (AMD, or NVidia) you should be able to use them simultaneously to render in Blender Cycles. If you keep them out of any proprietary GPU pairing technologies setup like SLI or Crossfire, the operating system detects them as two discrete GPUs and see both graphics cards, Blender should be able to see them as ...


11

The NVIDIA GeForce 9500 GT is from the GeForce 9 series, which doesn't fulfill the minimum requirements for Blender 2.8x. Supported Graphics Cards NVIDIA: GeForce 400 and newer, Quadro Tesla GPU architecture and newer, with NVIDIA drivers (list of all GeForce and Quadro GPUs) CUDA CUDA requires graphics cards with compute capability 3....


10

I'm quite surprised at the amount of false information here. First off Intel does indeed support OpenCL, as far as I can see up to 2.1. Not only this, but I have actual field experience with working with OpenCL on Intel APUs. The implementation works, and works well and will speed your processing the same way any other GPGPU compute technology would. ...


9

As iKlsR said in the comments to your question, we discarded the sm_13 cycles kernel... This is exactly the one needed for your series of cards. Please take my advice when I talk to you as a developer - you don't want to enable this for your card. First of all it's a hassle to setup nowadays! We don't care for this cards anymore, we don't even check if the ...


9

You will need to install the package nvidia-modprobe which (I believe) allows Blender to detect your nvidia CUDA device. This is applicable to Mint, Debian and Ubuntu, I don't know about other distributions of Linux: sudo apt-get install nvidia-modprobe You can also install nvidia-modprobe via your favourate package manager. Please use the Nvidia drivers ...


9

As it's still quite experimental and there are quite a few features still missing, it's only enabled for AMD gpus by default. It's possible to "force" it to work on other opencl devices by setting the environment variable CYCLES_OPENCL_SPLIT_KERNEL_TEST=1 Linux On Linux, run the following in a terminal: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1 blender ...


8

I have 3 cards, 2 cards scaled with a factor of 2.0, now the 3rd card scaled something along 1.7-1.9 factor. So 2 cards make a huge difference in cycles. You will for sure cut the render times in half. All articles I've read suggest not using SLI for apps using CUDA/Compute. If you're on Windows you can via drivers enable it or disable it. It's currently ...


8

I will assume you are talking about an external GPU in this case, say a Nvidia GTX 660 etc.. Excerpt from the Manual ...GPU rendering makes it possible to use your graphics card for rendering, instead of the CPU. This can speed up rendering, because modern GPUs are quite designed to do a lot of number crunching... In short, No, you do not need an external ...


8

The short answer is that SSS on the GPU eats up a lot of memory, so much so that it is recommended to have more than 1 GB of memory on for your GPU. This was mentioned in one of the videos from the Blender Conference (unfortunately I can't remember which one). Updating your drivers won't really help as that can't add more memory, so for now you are stuck ...


8

Go with the Nvidia. Performance aside, CUDA works better with Cycles. Until recently, OpenCL was not even supported in Cycles so it is not as bug-free and stable as CUDA. Another thing to consider is that usually new features are supported on just CPU first, then CUDA, then OpenCL. So be prepared to wait a little while to use new features on your GPU if ...


8

I think you've misunderstood the description for that build. It is merely stating what features were enabled at compile time. It's possible to build Blender without some things, such as the smoke simulator, or without CUDA. The description merely says that CUDA Cycles and the smoke simulator are both enabled, and that the build is speed optimized. It does ...


8

You cannot enable it since it is not yet supported. You will need to wait until it is, or attempt to build Blender with the support for Cuda 10 yourself apparently.


8

Unfortunately Apple has deprecated OpenCL and OpenGL in favor of their own Metal API. In fact, Blender has recently dropped OpenCL support on macOS completely, due to too many bugs in the macOS OpenCL compiler. Long story short, due to apple's rejection of widely used open APIs, it's unlikely there will be any support for OpenCL rendering on macOS in the ...


8

The Blender EEVEE render engine uses OpenGL for all the rendering, so YES, rendering on the CPU should be completely doable, as long as a viable CPU-based OpenGL implementation is also installed and enabled. In particular, Blender lists system requirements of only "OpenGL 3.3", which is officially supported by both softpipe and LLVMpipe. Base Knowledge ...


7

Quadro cards tend to have less CUDA cores than GTX cards, therefore their performance is worse. The advantage of using a Quadro or Tesla cards consists in more GPU memory (which means more complex renders). For more info you can check out the nVidia website and look at the specs of the cards.


7

So I just tested this: No, card order does not make a difference. Doesn't really matter which one is set as system (if the system is doing nothing) or to which the monitors are plugged. The render times were off by fractions of a second (BMW benchmark scene with 2 cars). Both my cards are in PCI Express 3.0 at x16 slots.


Only top voted, non community-wiki answers of a minimum length are eligible